flet ui (experimental)

Fix #152 : preview buttons on webUI

.github/FUNDING.yml

@@ -0,0 +1,15 @@
+# These are supported funding model platforms
+
+github: [jborza, jeremiahsb, mohangk]
+patreon: # Replace with a single Patreon username
+open_collective: # Replace with a single Open Collective username
+ko_fi: # Replace with a single Ko-fi username
+tidelift: # Replace with a single Tidelift platform-name/package-name e.g., npm/babel
+community_bridge: # Replace with a single Community Bridge project-name e.g., cloud-foundry
+liberapay: # Replace with a single Liberapay username
+issuehunt: # Replace with a single IssueHunt username
+lfx_crowdfunding: # Replace with a single LFX Crowdfunding project-name e.g., cloud-foundry
+polar: # Replace with a single Polar username
+buy_me_a_coffee: # Replace with a single Buy Me a Coffee username
+thanks_dev: # Replace with a single thanks.dev username
+custom: # Replace with up to 4 custom sponsorship URLs e.g., ['link1', 'link2']

.gitignore

@@ -38,3 +38,4 @@ dist/
 .old/
 test_assets/
 dev_notes/
+.claude/

abogen/VERSION

@@ -1 +1 @@
-1.3.0
+1.3.1

abogen/book_parser.py

@@ -915,7 +915,11 @@ class EpubParser(BaseBookParser):
 
             if slice_html.strip():
                 slice_soup = BeautifulSoup(slice_html, "html.parser")
-                for tag in slice_soup.find_all(["p", "div"]):
+
+                # Add line breaks after block-level elements to ensure pauses in speech
+                for tag in slice_soup.find_all(
+                    ["p", "div", "h1", "h2", "h3", "h4", "h5", "h6", "li", "blockquote"]
+                ):
                     tag.append("\n\n")
 
                 for ol in slice_soup.find_all("ol"):

abogen/conversion.py

@@ -1,16 +0,0 @@
-"""Backwards-compatible re-export of conversion module.
-
-The PyQt-based implementation lives in abogen.pyqt.conversion.
-The web-based implementation is in abogen.webui.conversion_runner.
-"""
-
-from __future__ import annotations
-
-# Re-export PyQt conversion classes for backwards compatibility
-from abogen.pyqt.conversion import (  # noqa: F401
-    ConversionThread,
-    VoicePreviewThread,
-    PlayAudioThread,
-)
-
-__all__ = ["ConversionThread", "VoicePreviewThread", "PlayAudioThread"]

abogen/frontend/__init__.py

@@ -0,0 +1,8 @@
+"""
+Abogen Flet Frontend Package.
+
+This package provides a unified, dual-target (desktop + web) user interface
+for the Abogen audiobook generation application, built with the Flet framework.
+"""
+
+__all__ = ["main"]

abogen/frontend/components/__init__.py

@@ -0,0 +1,32 @@
+"""Components sub-package."""
+from .widgets import (
+    resolve_icon,
+    build_drop_zone,
+    build_log_terminal,
+    log_entry,
+    build_progress_row,
+    build_primary_button,
+    build_secondary_button,
+    build_card,
+    build_section_header,
+    build_status_badge,
+    labelled_row,
+    show_snack,
+    build_divider,
+)
+
+__all__ = [
+    "build_drop_zone",
+    "resolve_icon",
+    "build_log_terminal",
+    "log_entry",
+    "build_progress_row",
+    "build_primary_button",
+    "build_secondary_button",
+    "build_card",
+    "build_section_header",
+    "build_status_badge",
+    "labelled_row",
+    "show_snack",
+    "build_divider",
+]

abogen/frontend/components/widgets.py

@@ -0,0 +1,630 @@
+"""
+Reusable UI components for the Abogen Flet frontend.
+
+Each function in this module returns a standalone Flet control or small
+widget tree.  Components read the current palette from the page's theme
+mode and should not hold any mutable state themselves – state lives in the
+session's ``AppState`` object.
+"""
+
+from __future__ import annotations
+
+from typing import Any, Callable, List, Optional
+
+import flet as ft
+
+from ..utils.theme import get_palette, RADIUS_MD, RADIUS_SM, SPACE_SM, SPACE_MD, SPACE_LG
+
+
+# ---------------------------------------------------------------------------
+# Helpers
+# ---------------------------------------------------------------------------
+
+
+def resolve_icon(icon: Any) -> Any:
+    """Convert a snake_case icon name to Flet IconData when possible."""
+    if isinstance(icon, str):
+        return getattr(ft.Icons, icon.upper(), icon)
+    return icon
+
+
+# ---------------------------------------------------------------------------
+# Drop-zone (file input area)
+# ---------------------------------------------------------------------------
+
+
+def build_drop_zone(
+    *,
+    on_pick: Callable[[], None],
+    label: str = "Drag & drop your file here or click to browse",
+    sub_label: str = "Supports: .txt · .epub · .pdf · .md · .srt · .ass · .vtt",
+    accent: bool = False,
+    error: bool = False,
+    filename: Optional[str] = None,
+    file_size: Optional[str] = None,
+    char_count: Optional[str] = None,
+    page: Optional[ft.Page] = None,
+) -> ft.GestureDetector:
+    """
+    Build an interactive file drop-zone widget.
+
+    The zone shows a dashed border and centred instructions by default,
+    switching to an 'active' green style when a file is loaded and a red
+    style when an error has occurred.
+
+    Args:
+        on_pick: Callback invoked when the user clicks or activates the zone.
+        label: Primary instruction text.
+        sub_label: Secondary hint text shown beneath the label.
+        accent: When True, renders the 'active/success' green style.
+        error: When True, renders the 'error/red' style.
+        filename: When provided, replaces the instruction text with file info.
+        file_size: Human-readable file size to display alongside the filename.
+        char_count: Character count to display alongside file info.
+        page: The current Flet ``Page``; used to derive the active palette.
+
+    Returns:
+        A ``ft.GestureDetector`` wrapping the visual drop-zone container.
+    """
+    dark = page is not None and page.theme_mode == ft.ThemeMode.DARK
+    p = get_palette(page) if page else None
+
+    # Colour scheme
+    if error:
+        border_color = "#e84e3c" if dark else "#c0392b"
+        bg_color = "#1a0a08" if dark else "#fff5f5"
+        text_color = "#e84e3c" if dark else "#c0392b"
+        icon_name = "error_outline"
+    elif accent:
+        border_color = "#42ad4a" if dark else "#2e9437"
+        bg_color = "#091810" if dark else "#f0fff1"
+        text_color = "#42ad4a" if dark else "#2e9437"
+        icon_name = "check_circle_outline"
+    else:
+        border_color = "#3a4466" if dark else "#a8b4d0"
+        bg_color = "#151928" if dark else "#f7f8fd"
+        text_color = "#9ba3b8" if dark else "#5a6172"
+        icon_name = "upload_file"
+
+    if filename:
+        # Compact file-info display
+        info_rows: List[ft.Control] = [
+            ft.Row(
+                [
+                    ft.Icon(resolve_icon("insert_drive_file"), color=text_color, size=28),
+                    ft.Column(
+                        [
+                            ft.Text(
+                                filename,
+                                weight=ft.FontWeight.W_600,
+                                size=13,
+                                color=text_color,
+                                no_wrap=False,
+                                max_lines=2,
+                                overflow=ft.TextOverflow.ELLIPSIS,
+                            ),
+                        ],
+                        tight=True,
+                        expand=True,
+                    ),
+                ],
+                alignment=ft.MainAxisAlignment.CENTER,
+                spacing=SPACE_SM,
+            )
+        ]
+        if file_size or char_count:
+            chips: List[ft.Control] = []
+            if file_size:
+                chips.append(
+                    ft.Text(f"📄 {file_size}", size=11, color=text_color, italic=True)
+                )
+            if char_count:
+                chips.append(
+                    ft.Text(f"🔤 {char_count} chars", size=11, color=text_color, italic=True)
+                )
+            info_rows.append(
+                ft.Row(chips, alignment=ft.MainAxisAlignment.CENTER, spacing=SPACE_MD)
+            )
+        content = ft.Column(
+            info_rows,
+            alignment=ft.MainAxisAlignment.CENTER,
+            horizontal_alignment=ft.CrossAxisAlignment.CENTER,
+            spacing=SPACE_SM,
+        )
+    else:
+        content = ft.Column(
+            [
+                ft.Icon(resolve_icon(icon_name), size=48, color=border_color, opacity=0.8),
+                ft.Text(
+                    label,
+                    size=14,
+                    weight=ft.FontWeight.W_500,
+                    color=text_color,
+                    text_align=ft.TextAlign.CENTER,
+                ),
+                ft.Text(
+                    sub_label,
+                    size=11,
+                    color=text_color,
+                    opacity=0.6,
+                    text_align=ft.TextAlign.CENTER,
+                ),
+            ],
+            alignment=ft.MainAxisAlignment.CENTER,
+            horizontal_alignment=ft.CrossAxisAlignment.CENTER,
+            spacing=SPACE_SM,
+        )
+
+    inner = ft.Container(
+        content=content,
+        border=ft.Border.all(2, border_color),
+        border_radius=RADIUS_MD,
+        bgcolor=bg_color,
+        padding=ft.Padding.all(SPACE_LG),
+        height=160,
+        alignment=ft.Alignment.CENTER,
+        expand=True,
+    )
+
+    return ft.GestureDetector(
+        content=ft.Row([inner], spacing=0),
+        on_tap=lambda _: on_pick(),
+        mouse_cursor=ft.MouseCursor.CLICK,
+    )
+
+
+# ---------------------------------------------------------------------------
+# Log terminal
+# ---------------------------------------------------------------------------
+
+
+def build_log_terminal(
+    *,
+    ref: Optional[ft.Ref] = None,
+    max_height: int = 260,
+    page: Optional[ft.Page] = None,
+) -> ft.Container:
+    """
+    Build a scrollable, read-only log terminal widget.
+
+    Args:
+        ref: Optional ``ft.Ref[ft.ListView]`` to bind the inner list-view so
+             callers can append entries programmatically.
+        max_height: Maximum pixel height before vertical scrolling activates.
+        page: Current Flet ``Page`` for palette derivation.
+
+    Returns:
+        A styled ``ft.Container`` wrapping a ``ft.ListView``.
+    """
+    dark = page is not None and page.theme_mode == ft.ThemeMode.DARK
+    bg = "#0d1117" if dark else "#f8f9fc"
+    text_color = "#b0b8cc" if dark else "#3d4358"
+    border_color = "#252a38" if dark else "#dce0ea"
+
+    list_view = ft.ListView(
+        expand=True,
+        auto_scroll=True,
+        spacing=1,
+        padding=ft.Padding.all(SPACE_SM),
+    )
+    if ref is not None:
+        ref.current = list_view
+
+    return ft.Container(
+        content=list_view,
+        bgcolor=bg,
+        border=ft.Border.all(1, border_color),
+        border_radius=RADIUS_SM,
+        height=max_height,
+        clip_behavior=ft.ClipBehavior.HARD_EDGE,
+    )
+
+
+def log_entry(message: str, level: str = "info", page: Optional[ft.Page] = None) -> ft.Text:
+    """
+    Create a single log-line ``ft.Text`` widget with appropriate colour coding.
+
+    Args:
+        message: The log message string.
+        level: Severity string: ``'info'``, ``'success'``, ``'error'``,
+               ``'warning'``, ``'debug'``, ``'critical'``.
+        page: Current Flet ``Page`` for dark/light mode detection.
+
+    Returns:
+        A styled ``ft.Text`` control.
+    """
+    dark = page is not None and page.theme_mode == ft.ThemeMode.DARK
+    palette: dict[str, str] = {
+        "info": "#9ba3b8" if dark else "#5a6172",
+        "success": "#42ad4a" if dark else "#2e9437",
+        "error": "#e84e3c" if dark else "#c0392b",
+        "warning": "#f5a623" if dark else "#d4870a",
+        "debug": "#5a6172" if dark else "#9ba3b8",
+        "critical": "#ff5722",
+        "trace": "#4e5568" if dark else "#b0b8cc",
+    }
+    color = palette.get(level.lower(), palette["info"])
+    return ft.Text(message, size=12, color=color, selectable=True, no_wrap=False)
+
+
+# ---------------------------------------------------------------------------
+# Progress row
+# ---------------------------------------------------------------------------
+
+
+def build_progress_row(
+    *,
+    progress_value: float = 0.0,
+    etr_text: str = "",
+    page: Optional[ft.Page] = None,
+) -> ft.Column:
+    """
+    Build a progress-bar + ETR-label column.
+
+    Args:
+        progress_value: Float in [0.0, 1.0].
+        etr_text: Pre-formatted estimated-time-remaining string.
+        page: Current ``Page`` for palette derivation.
+
+    Returns:
+        A ``ft.Column`` containing the progress bar and label.
+    """
+    dark = page is not None and page.theme_mode == ft.ThemeMode.DARK
+    fill = "#5b8af5" if dark else "#3a5fc4"
+    bg = "#1e2230" if dark else "#e4e8f0"
+
+    bar = ft.ProgressBar(
+        value=progress_value,
+        color=fill,
+        bgcolor=bg,
+        height=8,
+        border_radius=ft.BorderRadius.all(4),
+        expand=True,
+    )
+    label = ft.Text(
+        etr_text,
+        size=11,
+        color="#9ba3b8" if dark else "#5a6172",
+        text_align=ft.TextAlign.CENTER,
+    )
+    return ft.Column(
+        [bar, label],
+        spacing=SPACE_SM,
+        horizontal_alignment=ft.CrossAxisAlignment.CENTER,
+    )
+
+
+# ---------------------------------------------------------------------------
+# Primary action button
+# ---------------------------------------------------------------------------
+
+
+def build_primary_button(
+    text: str,
+    *,
+    icon: Optional[str] = None,
+    on_click: Optional[Callable] = None,
+    disabled: bool = False,
+    width: Optional[int] = None,
+    page: Optional[ft.Page] = None,
+) -> ft.ElevatedButton:
+    """
+    Build a prominent, styled primary action button.
+
+    Args:
+        text: Button label.
+        icon: Optional Flet icon name (e.g. ``'play_arrow'``).
+        on_click: Click callback.
+        disabled: Whether the button is non-interactive.
+        width: Optional fixed pixel width.
+        page: Current ``Page`` for accent colour derivation.
+
+    Returns:
+        A styled ``ft.ElevatedButton``.
+    """
+    dark = page is not None and page.theme_mode == ft.ThemeMode.DARK
+    bg = "#5b8af5" if dark else "#3a5fc4"
+    on_bg = "#ffffff"
+
+    style = ft.ButtonStyle(
+        bgcolor={
+            ft.ControlState.DEFAULT: bg,
+            ft.ControlState.HOVERED: "#3a5fc4" if dark else "#2a4fae",
+            ft.ControlState.DISABLED: "#2a2f3f" if dark else "#c0c8d8",
+        },
+        color={
+            ft.ControlState.DEFAULT: on_bg,
+            ft.ControlState.DISABLED: "#4e5568" if dark else "#9ba3b8",
+        },
+        elevation={"default": 2, "hovered": 4},
+        padding=ft.Padding.symmetric(horizontal=SPACE_LG, vertical=SPACE_MD),
+        shape=ft.RoundedRectangleBorder(radius=RADIUS_SM),
+        animation_duration=150,
+    )
+
+    return ft.ElevatedButton(
+        content=text,
+        icon=resolve_icon(icon),
+        on_click=on_click,
+        disabled=disabled,
+        width=width,
+        style=style,
+        height=48,
+    )
+
+
+# ---------------------------------------------------------------------------
+# Secondary / ghost button
+# ---------------------------------------------------------------------------
+
+
+def build_secondary_button(
+    text: str,
+    *,
+    icon: Optional[str] = None,
+    on_click: Optional[Callable] = None,
+    disabled: bool = False,
+    page: Optional[ft.Page] = None,
+) -> ft.OutlinedButton:
+    """
+    Build a secondary outlined button.
+
+    Args:
+        text: Button label.
+        icon: Optional Flet icon name.
+        on_click: Click callback.
+        disabled: Whether the button is non-interactive.
+        page: Current ``Page`` for border colour derivation.
+
+    Returns:
+        A styled ``ft.OutlinedButton``.
+    """
+    dark = page is not None and page.theme_mode == ft.ThemeMode.DARK
+    border_clr = "#3a4466" if dark else "#a8b4d0"
+    text_clr = "#e8eaf0" if dark else "#1a1d27"
+
+    style = ft.ButtonStyle(
+        side={
+            ft.ControlState.DEFAULT: ft.BorderSide(1.5, border_clr),
+            ft.ControlState.HOVERED: ft.BorderSide(1.5, "#5b8af5" if dark else "#3a5fc4"),
+        },
+        color={
+            ft.ControlState.DEFAULT: text_clr,
+            ft.ControlState.HOVERED: "#5b8af5" if dark else "#3a5fc4",
+            ft.ControlState.DISABLED: "#4e5568" if dark else "#9ba3b8",
+        },
+        padding=ft.Padding.symmetric(horizontal=SPACE_LG, vertical=SPACE_MD),
+        shape=ft.RoundedRectangleBorder(radius=RADIUS_SM),
+        animation_duration=150,
+    )
+
+    return ft.OutlinedButton(
+        content=text,
+        icon=resolve_icon(icon),
+        on_click=on_click,
+        disabled=disabled,
+        style=style,
+        height=44,
+    )
+
+
+# ---------------------------------------------------------------------------
+# Section card
+# ---------------------------------------------------------------------------
+
+
+def build_card(
+    content: ft.Control,
+    *,
+    padding: int = SPACE_LG,
+    page: Optional[ft.Page] = None,
+) -> ft.Container:
+    """
+    Wrap a control in a styled card container.
+
+    Args:
+        content: The child control to embed.
+        padding: Internal padding in pixels.
+        page: Current ``Page`` for palette derivation.
+
+    Returns:
+        A styled ``ft.Container``.
+    """
+    dark = page is not None and page.theme_mode == ft.ThemeMode.DARK
+    bg = "#181b23" if dark else "#ffffff"
+    border_clr = "#2c3147" if dark else "#dce0ea"
+
+    return ft.Container(
+        content=content,
+        bgcolor=bg,
+        border=ft.Border.all(1, border_clr),
+        border_radius=RADIUS_MD,
+        padding=ft.Padding.all(padding),
+        shadow=ft.BoxShadow(
+            spread_radius=0,
+            blur_radius=12,
+            color=ft.Colors.with_opacity(0.12 if dark else 0.06, ft.Colors.BLACK),
+            offset=ft.Offset(0, 2),
+        ),
+    )
+
+
+# ---------------------------------------------------------------------------
+# Section header
+# ---------------------------------------------------------------------------
+
+
+def build_section_header(
+    title: str,
+    *,
+    subtitle: Optional[str] = None,
+    icon: Optional[str] = None,
+    page: Optional[ft.Page] = None,
+) -> ft.Row:
+    """
+    Build a consistent section header row with an optional icon.
+
+    Args:
+        title: Section heading text.
+        subtitle: Optional explanatory sub-text.
+        icon: Optional Flet icon name.
+        page: Current ``Page`` for palette derivation.
+
+    Returns:
+        A ``ft.Row`` containing the icon and text column.
+    """
+    dark = page is not None and page.theme_mode == ft.ThemeMode.DARK
+    title_color = "#e8eaf0" if dark else "#1a1d27"
+    sub_color = "#9ba3b8" if dark else "#5a6172"
+    accent = "#5b8af5" if dark else "#3a5fc4"
+
+    children: List[ft.Control] = []
+    if icon:
+        children.append(ft.Icon(resolve_icon(icon), size=20, color=accent))
+
+    text_parts: List[ft.Control] = [
+        ft.Text(title, size=15, weight=ft.FontWeight.W_600, color=title_color)
+    ]
+    if subtitle:
+        text_parts.append(ft.Text(subtitle, size=11, color=sub_color))
+
+    children.append(
+        ft.Column(text_parts, spacing=1, tight=True, expand=True)
+    )
+
+    return ft.Row(children, spacing=SPACE_SM, vertical_alignment=ft.CrossAxisAlignment.START)
+
+
+# ---------------------------------------------------------------------------
+# Status badge
+# ---------------------------------------------------------------------------
+
+
+def build_status_badge(
+    label: str,
+    *,
+    variant: str = "info",
+    page: Optional[ft.Page] = None,
+) -> ft.Container:
+    """
+    Build a small status badge chip.
+
+    Args:
+        label: Badge text.
+        variant: Colour variant: ``'info'``, ``'success'``, ``'error'``,
+                 ``'warning'``, ``'neutral'``.
+        page: Current ``Page`` for theme derivation.
+
+    Returns:
+        A pill-shaped ``ft.Container``.
+    """
+    dark = page is not None and page.theme_mode == ft.ThemeMode.DARK
+    palette = {
+        "info":    ("#1a2a5e" if dark else "#dde8ff", "#5b8af5" if dark else "#3a5fc4"),
+        "success": ("#0d2010" if dark else "#d4f4d7", "#42ad4a" if dark else "#2e9437"),
+        "error":   ("#2a0a08" if dark else "#ffe0dc", "#e84e3c" if dark else "#c0392b"),
+        "warning": ("#2a1a00" if dark else "#fff4d8", "#f5a623" if dark else "#d4870a"),
+        "neutral": ("#1e2230" if dark else "#edf0f5", "#9ba3b8" if dark else "#5a6172"),
+    }
+    bg, fg = palette.get(variant, palette["info"])
+
+    return ft.Container(
+        content=ft.Text(label, size=10, weight=ft.FontWeight.W_600, color=fg),
+        bgcolor=bg,
+        border_radius=999,
+        padding=ft.Padding.symmetric(horizontal=8, vertical=3),
+    )
+
+
+# ---------------------------------------------------------------------------
+# Labelled control row
+# ---------------------------------------------------------------------------
+
+
+def labelled_row(
+    label: str,
+    control: ft.Control,
+    *,
+    label_width: int = 200,
+    tooltip: Optional[str] = None,
+    page: Optional[ft.Page] = None,
+) -> ft.Row:
+    """
+    Lay a label and a control side-by-side in a consistent row.
+
+    Args:
+        label: Human-readable label text.
+        control: The UI control placed to the right of the label.
+        label_width: Fixed pixel width of the label column.
+        tooltip: Optional tooltip text on the label.
+        page: Current ``Page`` for palette derivation.
+
+    Returns:
+        A ``ft.Row`` with the label pinned to a fixed width.
+    """
+    dark = page is not None and page.theme_mode == ft.ThemeMode.DARK
+    lbl_color = "#9ba3b8" if dark else "#5a6172"
+
+    lbl = ft.Text(label, size=13, color=lbl_color, weight=ft.FontWeight.W_500, width=label_width)
+    if tooltip:
+        lbl.tooltip = tooltip
+
+    return ft.Row(
+        [lbl, ft.Container(content=control, expand=True)],
+        alignment=ft.MainAxisAlignment.START,
+        vertical_alignment=ft.CrossAxisAlignment.CENTER,
+        spacing=SPACE_MD,
+    )
+
+
+# ---------------------------------------------------------------------------
+# Snack-bar helper
+# ---------------------------------------------------------------------------
+
+
+def show_snack(
+    page: ft.Page,
+    message: str,
+    *,
+    error: bool = False,
+    duration: int = 3000,
+) -> None:
+    """
+    Display a brief snack-bar notification.
+
+    Args:
+        page: The Flet ``Page`` instance.
+        message: Text to display.
+        error: When True, colours the bar red instead of the default accent.
+        duration: Visible duration in milliseconds.
+    """
+    dark = page.theme_mode == ft.ThemeMode.DARK
+    bg = "#e84e3c" if error else ("#5b8af5" if dark else "#3a5fc4")
+    page.snack_bar = ft.SnackBar(
+        content=ft.Text(message, color="#ffffff", size=13),
+        bgcolor=bg,
+        duration=duration,
+        show_close_icon=True,
+        close_icon_color="#ffffff",
+    )
+    page.snack_bar.open = True
+    page.update()
+
+
+# ---------------------------------------------------------------------------
+# Divider helper
+# ---------------------------------------------------------------------------
+
+
+def build_divider(page: Optional[ft.Page] = None) -> ft.Divider:
+    """
+    Build a styled horizontal rule divider.
+
+    Args:
+        page: Current ``Page`` for palette derivation.
+
+    Returns:
+        A ``ft.Divider``.
+    """
+    dark = page is not None and page.theme_mode == ft.ThemeMode.DARK
+    return ft.Divider(color="#252a38" if dark else "#e8ebf2", height=1, thickness=1)

abogen/frontend/main.py

@@ -0,0 +1,365 @@
+"""
+Abogen Flet Frontend – main entry point.
+
+Run as desktop app:
+    python -m abogen.frontend.main
+
+Run as web app (binds to port 8080 by default):
+    python -m abogen.frontend.main --web --port 8080
+
+Architecture
+------------
+One ``ft.app()`` call launches the server.  For every new browser tab (or the
+desktop window) Flet invokes ``_app_entry(page)`` in its own coroutine, which
+creates a fresh ``AppState`` and wires together the navigation rail and views.
+This guarantees complete per-session isolation in multi-user web deployments.
+"""
+from __future__ import annotations
+
+import argparse
+import sys
+from pathlib import Path
+from typing import Optional
+
+import flet as ft
+
+from .state import AppState
+from .components import resolve_icon
+from .views.dashboard import DashboardView
+from .views.settings import SettingsView
+from .views.queue_view import QueueView
+from .utils.theme import make_theme, DARK, LIGHT, SPACE_SM, SPACE_MD, SPACE_LG, RADIUS_MD
+from abogen.constants import PROGRAM_NAME as APP_NAME
+
+
+# ---------------------------------------------------------------------------
+# Navigation destinations
+# ---------------------------------------------------------------------------
+
+_NAV_ITEMS = [
+    ("Convert", "swap_horiz", "swap_horiz"),
+    ("Queue",   "list_alt",   "list_alt"),
+    ("Settings", "settings",  "settings"),
+]
+
+_ASSETS_DIR = Path(__file__).resolve().parents[1] / "assets"
+
+
+def _build_sidebar_item(
+    *,
+    label: str,
+    icon: str,
+    selected: bool,
+    palette,
+    on_click,
+) -> ft.Container:
+    accent = palette.accent if selected else palette.text_secondary
+    bg = palette.sidebar_selected_bg if selected else palette.sidebar_bg
+    return ft.Container(
+        content=ft.Row(
+            [
+                ft.Icon(resolve_icon(icon), size=20, color=accent),
+                ft.Text(
+                    label,
+                    size=13,
+                    weight=ft.FontWeight.W_600 if selected else ft.FontWeight.W_500,
+                    color=accent,
+                ),
+            ],
+            spacing=SPACE_MD,
+            vertical_alignment=ft.CrossAxisAlignment.CENTER,
+        ),
+        bgcolor=bg,
+        border_radius=RADIUS_MD,
+        padding=ft.Padding.symmetric(horizontal=SPACE_MD, vertical=10),
+        ink=True,
+        on_click=on_click,
+    )
+
+
+# ---------------------------------------------------------------------------
+# Per-session entry point
+# ---------------------------------------------------------------------------
+
+def _app_entry(page: ft.Page) -> None:
+    try:
+        # ── State ────────────────────────────────────────────────────────────
+        state = AppState()
+        state.load_from_config()
+
+        # ── Page basics ──────────────────────────────────────────────────────
+        page.title = APP_NAME
+        page.padding = 0
+        page.spacing = 0
+        page.bgcolor = DARK.bg_base
+        page.theme_mode = ft.ThemeMode.DARK
+        page.theme = make_theme(dark=True)
+        page.dark_theme = make_theme(dark=True)
+        page.fonts = {}
+        page.window.min_width = 520
+        page.window.min_height = 600
+        page.update()
+
+        # ── Content area ref ─────────────────────────────────────────────────
+        content_area = ft.Column(expand=True, spacing=0)
+        sidebar_body = ft.Column(spacing=SPACE_SM)
+        theme_button_host = ft.Container()
+        brand_title = ft.Text(
+            APP_NAME,
+            size=18,
+            weight=ft.FontWeight.W_700,
+            color=DARK.text_primary,
+        )
+        brand_fallback_icon = ft.Icon(resolve_icon("speaker_notes"), size=32, color=DARK.accent)
+        divider = ft.VerticalDivider(width=1, color=DARK.border)
+
+        # ── Views ────────────────────────────────────────────────────────────
+        dashboard_view = DashboardView(page, state)
+        settings_view = SettingsView(page, state)
+        queue_view = QueueView(page, state)
+
+        views = [
+            dashboard_view.build,
+            queue_view.build,
+            settings_view.build,
+        ]
+        _selected_index = [0]
+
+        def _refresh_sidebar() -> None:
+            dark = page.theme_mode == ft.ThemeMode.DARK
+            pal = DARK if dark else LIGHT
+            sidebar_body.controls = [
+                _build_sidebar_item(
+                    label=label,
+                    icon=icon,
+                    selected=index == _selected_index[0],
+                    palette=pal,
+                    on_click=lambda _, i=index: _navigate(i),
+                )
+                for index, (label, icon, _) in enumerate(_NAV_ITEMS)
+            ]
+            sidebar.bgcolor = pal.sidebar_bg
+            divider.color = pal.border
+            brand_title.color = pal.text_primary
+            brand_fallback_icon.color = pal.accent
+            theme_button_host.content = ft.Container(
+                content=ft.Icon(
+                    resolve_icon("dark_mode" if dark else "light_mode"),
+                    size=20,
+                    color=pal.text_secondary,
+                ),
+                tooltip="Toggle theme",
+                border_radius=RADIUS_MD,
+                padding=8,
+                ink=True,
+                on_click=lambda _: _toggle_theme(page, _refresh_sidebar),
+            )
+
+        def _navigate(index: int) -> None:
+            _selected_index[0] = index
+            content_area.controls.clear()
+            built = views[index]()
+            content_area.controls.append(
+                ft.Container(
+                    content=built,
+                    expand=True,
+                    padding=ft.Padding.symmetric(horizontal=SPACE_LG, vertical=SPACE_LG),
+                )
+            )
+            _refresh_sidebar()
+            page.update()
+
+        # ── Sidebar ──────────────────────────────────────────────────────────
+        pal = DARK
+        sidebar = ft.Container(
+            width=220,
+            bgcolor=pal.sidebar_bg,
+            padding=ft.Padding.all(SPACE_MD),
+            content=ft.Column(
+                [
+                    ft.Container(
+                        content=ft.Row(
+                            [
+                                ft.Image(
+                                    src="icon.png",
+                                    width=36,
+                                    height=36,
+                                    fit=ft.BoxFit.CONTAIN,
+                                    error_content=brand_fallback_icon,
+                                ),
+                                brand_title,
+                            ],
+                            spacing=SPACE_MD,
+                            vertical_alignment=ft.CrossAxisAlignment.CENTER,
+                        ),
+                        padding=ft.Padding.only(top=SPACE_SM, bottom=SPACE_LG),
+                    ),
+                    sidebar_body,
+                    ft.Container(expand=True),
+                    ft.Row([theme_button_host], alignment=ft.MainAxisAlignment.END),
+                ],
+                expand=True,
+                spacing=SPACE_SM,
+            ),
+        )
+        _refresh_sidebar()
+
+        # ── Page handle for pubsub (queue → dashboard) ───────────────────────
+        def _handle_pubsub(topic: str) -> None:
+            if topic == "start_queue":
+                _navigate(0)
+
+        page.pubsub.subscribe(_handle_pubsub)
+
+        # ── Layout ────────────────────────────────────────────────────────────
+        page.add(
+            ft.Row(
+                [
+                    sidebar,
+                    divider,
+                    ft.Container(content=content_area, expand=True),
+                ],
+                expand=True,
+                spacing=0,
+                vertical_alignment=ft.CrossAxisAlignment.START,
+            )
+        )
+
+        # Show dashboard by default
+        _navigate(0)
+        page.update()
+    except Exception as e:
+        import traceback
+        traceback.print_exc()
+        print(f"ERROR IN _app_entry: {e}")
+        raise
+
+
+def _toggle_theme(page: ft.Page, refresh_sidebar) -> None:
+    """Switch between dark and light theme modes."""
+    if page.theme_mode == ft.ThemeMode.DARK:
+        page.theme_mode = ft.ThemeMode.LIGHT
+        page.bgcolor = LIGHT.bg_base
+    else:
+        page.theme_mode = ft.ThemeMode.DARK
+        page.bgcolor = DARK.bg_base
+
+    page.theme = make_theme(page.theme_mode == ft.ThemeMode.DARK)
+    refresh_sidebar()
+    page.update()
+
+
+# ---------------------------------------------------------------------------
+# CLI helpers & entry point
+# ---------------------------------------------------------------------------
+
+def _is_port_free(host: str, port: int) -> bool:
+    import socket
+    try:
+        with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s:
+            s.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
+            s.bind((host, port))
+            return True
+    except OSError:
+        return False
+
+
+def _find_free_port(host: str, start_port: int) -> int:
+    import socket
+    port = start_port
+    while port < 65535:
+        try:
+            with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s:
+                s.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
+                s.bind((host, port))
+                return port
+        except OSError:
+            port += 1
+    return start_port
+
+
+def main() -> None:
+    """
+    Start the Abogen Flet frontend.
+
+    Parses ``--web`` and ``--port`` CLI arguments to choose desktop vs. web
+    mode, then hands control to ``ft.app()``.
+    """
+    import logging
+    logging.basicConfig(level=logging.INFO)
+    logging.getLogger("flet").setLevel(logging.INFO)
+
+    parser = argparse.ArgumentParser(description=f"{APP_NAME} – Flet frontend")
+    parser.add_argument(
+        "--web", action="store_true",
+        help="Run as a web server instead of a desktop window.",
+    )
+    parser.add_argument(
+        "--port", type=int, default=8080,
+        help="Port for the web server (default: 8080). Ignored in desktop mode.",
+    )
+    parser.add_argument(
+        "--host", default="127.0.0.1",
+        help="Host for the web server (default: 127.0.0.1). Use 0.0.0.0 to expose publicly.",
+    )
+    args = parser.parse_args()
+
+    if args.web:
+        port_specified = "--port" in sys.argv
+        target_port = args.port
+
+        if not port_specified:
+            target_port = _find_free_port(args.host, 8080)
+            if target_port != 8080:
+                print(f"Port 8080 is in use. Automatically routed to free port: {target_port}")
+        else:
+            if not _is_port_free(args.host, target_port):
+                print(f"Error: Port {target_port} is already in use on {args.host}.", file=sys.stderr)
+                print("Please select a different port or omit the --port flag to find one automatically.", file=sys.stderr)
+                sys.exit(1)
+
+        print(f"Starting Abogen WebUI on http://{args.host}:{target_port} ...")
+        ft.app(
+            target=_app_entry,
+            view=ft.AppView.WEB_BROWSER,
+            port=target_port,
+            host=args.host,
+            assets_dir=str(_ASSETS_DIR) if _ASSETS_DIR.exists() else None,
+            no_cdn=True,
+            web_renderer="canvaskit",
+        )
+    else:
+        try:
+            ft.app(
+                target=_app_entry,
+                view=ft.AppView.FLET_APP,
+                assets_dir=str(_ASSETS_DIR) if _ASSETS_DIR.exists() else None,
+            )
+        except Exception as e:
+            print(f"Warning: Failed to launch native desktop window: {e}", file=sys.stderr)
+            print("Falling back to running as a web application in your default browser...", file=sys.stderr)
+            target_port = _find_free_port("127.0.0.1", 8080)
+            print(f"Starting Abogen WebUI on http://127.0.0.1:{target_port} ...")
+            ft.app(
+                target=_app_entry,
+                view=ft.AppView.WEB_BROWSER,
+                port=target_port,
+                host="127.0.0.1",
+                assets_dir=str(_ASSETS_DIR) if _ASSETS_DIR.exists() else None,
+                no_cdn=True,
+                web_renderer="canvaskit",
+            )
+
+
+def main_web() -> None:
+    """
+    Start the Abogen Flet frontend as a web server.
+    """
+    import sys
+    if "--web" not in sys.argv:
+        sys.argv.insert(1, "--web")
+    main()
+
+
+if __name__ == "__main__":
+    main()

abogen/frontend/state/__init__.py

@@ -0,0 +1,4 @@
+"""State sub-package – exports AppState and ConversionJob."""
+from .app_state import AppState, ConversionJob
+
+__all__ = ["AppState", "ConversionJob"]

abogen/frontend/state/app_state.py

@@ -0,0 +1,451 @@
+"""
+Centralized, per-session application state for the Abogen Flet frontend.
+
+Each Flet page (session) gets its own instance of AppState, which guarantees
+complete isolation between simultaneous web-browser clients and the desktop
+window. The class carries every configuration variable, file buffer reference,
+and generation progress field that the rest of the UI reads or writes.
+
+This module intentionally has no Flet imports so it can be unit-tested without
+a running Flet server.
+"""
+
+from __future__ import annotations
+
+import threading
+from dataclasses import dataclass, field
+from pathlib import Path
+from typing import Any, Callable, Dict, List, Optional
+
+from abogen.utils import load_config, save_config
+
+
+# ---------------------------------------------------------------------------
+# Helpers
+# ---------------------------------------------------------------------------
+
+
+def _default_config() -> Dict[str, Any]:
+    """Load the persisted user config dict, returning an empty dict on failure."""
+    try:
+        return load_config() or {}
+    except Exception:
+        return {}
+
+
+# ---------------------------------------------------------------------------
+# Per-session state
+# ---------------------------------------------------------------------------
+
+
+@dataclass
+class ConversionJob:
+    """Lightweight descriptor of a single queued conversion job."""
+
+    file_path: str
+    """Absolute path to the text/epub/pdf/txt input file."""
+
+    display_name: str
+    """User-visible filename (may be the original epub/pdf path)."""
+
+    voice: str
+    """Voice formula string (e.g. 'af_heart' or 'af_heart*0.5+am_adam*0.5')."""
+
+    lang_code: str
+    """Single-char language prefix used by Kokoro (e.g. 'a', 'b', 'e')."""
+
+    speed: float = 1.0
+    """Playback speed multiplier, range 0.1 – 2.0."""
+
+    output_format: str = "mp3"
+    """Output audio container format."""
+
+    subtitle_mode: str = "Disabled"
+    """Subtitle generation mode."""
+
+    save_option: str = "Save next to input file"
+    """Save location strategy."""
+
+    output_folder: Optional[str] = None
+    """Absolute path when save_option is 'Choose output folder'."""
+
+    char_count: int = 0
+    """Pre-computed character count for ETR estimation."""
+
+    replace_single_newlines: bool = True
+    save_chapters_separately: Optional[bool] = None
+    merge_chapters_at_end: Optional[bool] = None
+
+
+@dataclass
+class AppState:
+    """
+    Single source of truth for one Flet session.
+
+    Instantiated once per ``ft.app()`` call on desktop, and once per browser
+    tab on web.  All UI components receive a reference to this object and
+    read/write it to keep themselves in sync.
+
+    Thread-safety: mutation from background threads should be done via the
+    provided ``_lock``.  The UI update callbacks (``on_log``,
+    ``on_progress``, etc.) are always invoked on the Flet event loop via
+    ``page.run_task()`` and must be set by the view layer.
+    """
+
+    # -----------------------------------------------------------------------
+    # Runtime identity
+    # -----------------------------------------------------------------------
+    _lock: threading.Lock = field(default_factory=threading.Lock, repr=False, compare=False)
+
+    # -----------------------------------------------------------------------
+    # Persisted user config (loaded once, written on every change)
+    # -----------------------------------------------------------------------
+    config: Dict[str, Any] = field(default_factory=_default_config)
+
+    # -----------------------------------------------------------------------
+    # File / input state
+    # -----------------------------------------------------------------------
+    selected_file: Optional[str] = None
+    """Path to the processed text file (may be a temp cache copy for epub/pdf)."""
+
+    selected_file_type: Optional[str] = None
+    """'txt' | 'epub' | 'pdf' | 'markdown' | None"""
+
+    selected_book_path: Optional[str] = None
+    """Original epub/pdf path before being converted to txt."""
+
+    displayed_file_path: Optional[str] = None
+    """Path shown in the UI drop-zone (original book or txt file)."""
+
+    selected_chapters: List[str] = field(default_factory=list)
+    """Ordered list of selected chapter href tokens (or page numbers for PDFs)."""
+
+    save_chapters_separately: Optional[bool] = None
+    merge_chapters_at_end: Optional[bool] = None
+    save_as_project: bool = False
+    char_count: int = 0
+
+    # -----------------------------------------------------------------------
+    # Voice / language
+    # -----------------------------------------------------------------------
+    selected_voice: str = "af_heart"
+    selected_lang: str = "a"
+    selected_profile_name: Optional[str] = None
+    mixed_voice_state: Optional[List[Any]] = None
+    """List of [voice_id, weight] pairs when the formula mixer is in use."""
+
+    # -----------------------------------------------------------------------
+    # Conversion parameters
+    # -----------------------------------------------------------------------
+    speed: float = 1.0
+    use_gpu: bool = True
+    selected_format: str = "wav"
+    subtitle_mode: str = "Sentence"
+    subtitle_format: str = "ass_centered_narrow"
+    replace_single_newlines: bool = True
+    save_option: str = "Save next to input file"
+    selected_output_folder: Optional[str] = None
+    silence_duration: float = 2.0
+    max_subtitle_words: int = 50
+    separate_chapters_format: str = "wav"
+    use_silent_gaps: bool = True
+    subtitle_speed_method: str = "tts"
+    use_spacy_segmentation: bool = True
+    chunk_level: str = "paragraph"
+    generate_epub3: bool = False
+
+    # TTS provider
+    tts_provider: str = "kokoro"
+    supertonic_total_steps: int = 5
+
+    # Chapter options
+    chapter_intro_delay: float = 0.5
+    read_title_intro: bool = False
+    read_closing_outro: bool = True
+    auto_prefix_chapter_titles: bool = True
+    normalize_chapter_opening_caps: bool = True
+
+    # Speaker analysis
+    speaker_analysis_threshold: int = 3
+
+    # Word substitutions
+    word_substitutions_enabled: bool = False
+    word_substitutions_list: str = ""
+    case_sensitive_substitutions: bool = False
+    replace_all_caps: bool = False
+    replace_numerals: bool = False
+    fix_nonstandard_punctuation: bool = False
+
+    # -----------------------------------------------------------------------
+    # Conversion runtime state
+    # -----------------------------------------------------------------------
+    is_converting: bool = False
+    is_cancelled: bool = False
+    progress: float = 0.0
+    """Fractional progress 0.0 – 1.0."""
+    etr_seconds: Optional[float] = None
+    """Estimated seconds remaining, or None if unknown."""
+    last_output_path: Optional[str] = None
+    log_lines: List[str] = field(default_factory=list)
+    """Buffered log messages, capped at LOG_MAX_LINES."""
+
+    LOG_MAX_LINES: int = 2000
+
+    # -----------------------------------------------------------------------
+    # Queue
+    # -----------------------------------------------------------------------
+    queued_items: List[ConversionJob] = field(default_factory=list)
+    current_queue_index: int = 0
+
+    # -----------------------------------------------------------------------
+    # Callbacks (set by the view layer, not serialised)
+    # -----------------------------------------------------------------------
+    on_log: Optional[Callable[[str, str], None]] = field(default=None, repr=False, compare=False)
+    """Called from any thread: ``on_log(message, level)``."""
+
+    on_progress: Optional[Callable[[float, Optional[float]], None]] = field(
+        default=None, repr=False, compare=False
+    )
+    """Called from any thread: ``on_progress(fraction, etr_seconds)``."""
+
+    on_conversion_finished: Optional[Callable[[str, Optional[str]], None]] = field(
+        default=None, repr=False, compare=False
+    )
+    """Called from any thread: ``on_conversion_finished(message, output_path)``."""
+
+    # -----------------------------------------------------------------------
+    # Integrations
+    # -----------------------------------------------------------------------
+    audiobookshelf_enabled: bool = False
+    audiobookshelf_base_url: str = ""
+    audiobookshelf_api_token: str = ""
+    audiobookshelf_library_id: str = ""
+    audiobookshelf_folder_id: str = ""
+    audiobookshelf_verify_ssl: bool = True
+    audiobookshelf_auto_send: bool = False
+    audiobookshelf_send_cover: bool = True
+    audiobookshelf_send_chapters: bool = True
+    audiobookshelf_send_subtitles: bool = False
+    audiobookshelf_timeout: float = 30.0
+
+    calibre_opds_enabled: bool = False
+    calibre_opds_base_url: str = ""
+    calibre_opds_username: str = ""
+    calibre_opds_password: str = ""
+    calibre_opds_verify_ssl: bool = True
+
+    # -----------------------------------------------------------------------
+    # Public helpers
+    # -----------------------------------------------------------------------
+
+    def load_from_config(self) -> None:
+        """
+        Populate all fields from the persisted JSON config file.
+
+        Called once at startup and whenever the settings page is saved.
+        Thread-safe.
+        """
+        with self._lock:
+            cfg = _default_config()
+            self.config = cfg
+
+            self.selected_voice = cfg.get("selected_voice", "af_heart")
+            self.selected_lang = self.selected_voice[0] if self.selected_voice else "a"
+            self.selected_profile_name = cfg.get("selected_profile_name")
+            self.speed = cfg.get("speed", 1.0)
+            self.use_gpu = cfg.get("use_gpu", True)
+            self.selected_format = cfg.get("selected_format", "wav")
+            self.subtitle_mode = cfg.get("subtitle_mode", "Sentence")
+            self.subtitle_format = cfg.get("subtitle_format", "ass_centered_narrow")
+            self.replace_single_newlines = cfg.get("replace_single_newlines", True)
+            self.save_option = cfg.get("save_option", "Save next to input file")
+            self.selected_output_folder = cfg.get("selected_output_folder")
+            self.silence_duration = cfg.get("silence_duration", 2.0)
+            self.max_subtitle_words = cfg.get("max_subtitle_words", 50)
+            self.separate_chapters_format = cfg.get("separate_chapters_format", "wav")
+            self.use_silent_gaps = cfg.get("use_silent_gaps", True)
+            self.subtitle_speed_method = cfg.get("subtitle_speed_method", "tts")
+            self.use_spacy_segmentation = cfg.get("use_spacy_segmentation", True)
+            self.chunk_level = cfg.get("chunk_level", "paragraph")
+            self.generate_epub3 = cfg.get("generate_epub3", False)
+            self.tts_provider = cfg.get("tts_provider", "kokoro")
+            self.supertonic_total_steps = cfg.get("supertonic_total_steps", 5)
+            self.chapter_intro_delay = cfg.get("chapter_intro_delay", 0.5)
+            self.read_title_intro = cfg.get("read_title_intro", False)
+            self.read_closing_outro = cfg.get("read_closing_outro", True)
+            self.auto_prefix_chapter_titles = cfg.get("auto_prefix_chapter_titles", True)
+            self.normalize_chapter_opening_caps = cfg.get("normalize_chapter_opening_caps", True)
+            self.speaker_analysis_threshold = cfg.get("speaker_analysis_threshold", 3)
+            self.word_substitutions_enabled = cfg.get("word_substitutions_enabled", False)
+            self.word_substitutions_list = cfg.get("word_substitutions_list", "")
+            self.case_sensitive_substitutions = cfg.get("case_sensitive_substitutions", False)
+            self.replace_all_caps = cfg.get("replace_all_caps", False)
+            self.replace_numerals = cfg.get("replace_numerals", False)
+            self.fix_nonstandard_punctuation = cfg.get("fix_nonstandard_punctuation", False)
+
+            # Integrations
+            integrations: Dict[str, Any] = cfg.get("integrations", {})
+            abs_cfg = integrations.get("audiobookshelf", {})
+            self.audiobookshelf_enabled = bool(abs_cfg.get("enabled", False))
+            self.audiobookshelf_base_url = str(abs_cfg.get("base_url", ""))
+            self.audiobookshelf_api_token = str(abs_cfg.get("api_token", ""))
+            self.audiobookshelf_library_id = str(abs_cfg.get("library_id", ""))
+            self.audiobookshelf_folder_id = str(abs_cfg.get("folder_id", ""))
+            self.audiobookshelf_verify_ssl = bool(abs_cfg.get("verify_ssl", True))
+            self.audiobookshelf_auto_send = bool(abs_cfg.get("auto_send", False))
+            self.audiobookshelf_send_cover = bool(abs_cfg.get("send_cover", True))
+            self.audiobookshelf_send_chapters = bool(abs_cfg.get("send_chapters", True))
+            self.audiobookshelf_send_subtitles = bool(abs_cfg.get("send_subtitles", False))
+            self.audiobookshelf_timeout = float(abs_cfg.get("timeout", 30.0))
+
+            cal_cfg = integrations.get("calibre_opds", {})
+            self.calibre_opds_enabled = bool(cal_cfg.get("enabled", False))
+            self.calibre_opds_base_url = str(cal_cfg.get("base_url", ""))
+            self.calibre_opds_username = str(cal_cfg.get("username", ""))
+            self.calibre_opds_password = str(cal_cfg.get("password", ""))
+            self.calibre_opds_verify_ssl = bool(cal_cfg.get("verify_ssl", True))
+
+    def persist_config(self) -> None:
+        """
+        Write the current config snapshot back to disk.
+
+        Only the fields that map to the JSON config are written; runtime state
+        (progress, log_lines, callbacks) is not persisted.
+        Thread-safe.
+        """
+        with self._lock:
+            cfg = self.config.copy()
+            cfg["selected_voice"] = self.selected_voice
+            cfg["selected_profile_name"] = self.selected_profile_name
+            cfg["speed"] = self.speed
+            cfg["use_gpu"] = self.use_gpu
+            cfg["selected_format"] = self.selected_format
+            cfg["subtitle_mode"] = self.subtitle_mode
+            cfg["subtitle_format"] = self.subtitle_format
+            cfg["replace_single_newlines"] = self.replace_single_newlines
+            cfg["save_option"] = self.save_option
+            cfg["selected_output_folder"] = self.selected_output_folder
+            cfg["silence_duration"] = self.silence_duration
+            cfg["max_subtitle_words"] = self.max_subtitle_words
+            cfg["separate_chapters_format"] = self.separate_chapters_format
+            cfg["use_silent_gaps"] = self.use_silent_gaps
+            cfg["subtitle_speed_method"] = self.subtitle_speed_method
+            cfg["use_spacy_segmentation"] = self.use_spacy_segmentation
+            cfg["chunk_level"] = self.chunk_level
+            cfg["generate_epub3"] = self.generate_epub3
+            cfg["tts_provider"] = self.tts_provider
+            cfg["supertonic_total_steps"] = self.supertonic_total_steps
+            cfg["chapter_intro_delay"] = self.chapter_intro_delay
+            cfg["read_title_intro"] = self.read_title_intro
+            cfg["read_closing_outro"] = self.read_closing_outro
+            cfg["auto_prefix_chapter_titles"] = self.auto_prefix_chapter_titles
+            cfg["normalize_chapter_opening_caps"] = self.normalize_chapter_opening_caps
+            cfg["speaker_analysis_threshold"] = self.speaker_analysis_threshold
+            cfg["word_substitutions_enabled"] = self.word_substitutions_enabled
+            cfg["word_substitutions_list"] = self.word_substitutions_list
+            cfg["case_sensitive_substitutions"] = self.case_sensitive_substitutions
+            cfg["replace_all_caps"] = self.replace_all_caps
+            cfg["replace_numerals"] = self.replace_numerals
+            cfg["fix_nonstandard_punctuation"] = self.fix_nonstandard_punctuation
+            # Integrations
+            cfg.setdefault("integrations", {})
+            cfg["integrations"]["audiobookshelf"] = {
+                "enabled": self.audiobookshelf_enabled,
+                "base_url": self.audiobookshelf_base_url,
+                "api_token": self.audiobookshelf_api_token,
+                "library_id": self.audiobookshelf_library_id,
+                "folder_id": self.audiobookshelf_folder_id,
+                "verify_ssl": self.audiobookshelf_verify_ssl,
+                "auto_send": self.audiobookshelf_auto_send,
+                "send_cover": self.audiobookshelf_send_cover,
+                "send_chapters": self.audiobookshelf_send_chapters,
+                "send_subtitles": self.audiobookshelf_send_subtitles,
+                "timeout": self.audiobookshelf_timeout,
+            }
+            cfg["integrations"]["calibre_opds"] = {
+                "enabled": self.calibre_opds_enabled,
+                "base_url": self.calibre_opds_base_url,
+                "username": self.calibre_opds_username,
+                "password": self.calibre_opds_password,
+                "verify_ssl": self.calibre_opds_verify_ssl,
+            }
+            self.config = cfg
+            try:
+                save_config(cfg)
+            except Exception:
+                pass
+
+    def append_log(self, message: str, level: str = "info") -> None:
+        """
+        Thread-safely append a log line and trigger the UI callback.
+
+        Caps the internal buffer at ``LOG_MAX_LINES`` to prevent unbounded
+        memory growth during very long conversion tasks.
+        """
+        with self._lock:
+            self.log_lines.append(f"[{level.upper()}] {message}")
+            if len(self.log_lines) > self.LOG_MAX_LINES:
+                # Trim oldest 10 % to amortise the cost of trimming
+                trim = self.LOG_MAX_LINES // 10
+                self.log_lines = self.log_lines[trim:]
+
+        cb = self.on_log
+        if cb is not None:
+            try:
+                cb(message, level)
+            except Exception:
+                pass
+
+    def update_progress(self, fraction: float, etr: Optional[float] = None) -> None:
+        """
+        Update fractional progress and ETR, then notify the UI callback.
+
+        Args:
+            fraction: Value in [0.0, 1.0].
+            etr: Estimated seconds remaining, or None.
+        """
+        with self._lock:
+            self.progress = max(0.0, min(1.0, fraction))
+            self.etr_seconds = etr
+
+        cb = self.on_progress
+        if cb is not None:
+            try:
+                cb(fraction, etr)
+            except Exception:
+                pass
+
+    def get_voice_formula(self) -> str:
+        """
+        Return the effective voice formula string.
+
+        Uses the mixed_voice_state if the formula mixer is active, otherwise
+        returns the raw selected_voice.
+        """
+        if self.mixed_voice_state:
+            parts = [f"{name}*{weight}" for name, weight in self.mixed_voice_state]
+            return " + ".join(filter(None, parts))
+        return self.selected_voice or "af_heart"
+
+    def reset_file_state(self) -> None:
+        """Clear all file-related fields without touching voice/settings."""
+        with self._lock:
+            self.selected_file = None
+            self.selected_file_type = None
+            self.selected_book_path = None
+            self.displayed_file_path = None
+            self.selected_chapters = []
+            self.save_chapters_separately = None
+            self.merge_chapters_at_end = None
+            self.save_as_project = False
+            self.char_count = 0
+
+    def reset_conversion_state(self) -> None:
+        """Clear all runtime conversion fields to start fresh."""
+        with self._lock:
+            self.is_converting = False
+            self.is_cancelled = False
+            self.progress = 0.0
+            self.etr_seconds = None
+            self.last_output_path = None
+            self.log_lines = []

abogen/frontend/utils/__init__.py

@@ -0,0 +1,38 @@
+"""Utils sub-package."""
+from .helpers import (
+    human_readable_size,
+    format_duration,
+    format_etr,
+    detect_file_type,
+    is_supported_file,
+    is_book_type,
+    voice_lang_code,
+    language_label,
+    grouped_voices,
+    voice_display_name,
+    parse_voice_formula,
+    format_number,
+    safe_basename,
+    output_format_label,
+    subtitle_format_label,
+    SUPPORTED_EXTENSIONS,
+)
+
+__all__ = [
+    "human_readable_size",
+    "format_duration",
+    "format_etr",
+    "detect_file_type",
+    "is_supported_file",
+    "is_book_type",
+    "voice_lang_code",
+    "language_label",
+    "grouped_voices",
+    "voice_display_name",
+    "parse_voice_formula",
+    "format_number",
+    "safe_basename",
+    "output_format_label",
+    "subtitle_format_label",
+    "SUPPORTED_EXTENSIONS",
+]

abogen/frontend/utils/conversion_bridge.py

@@ -0,0 +1,462 @@
+"""
+Background conversion bridge for the Abogen Flet frontend.
+
+This module wraps the existing ``abogen.webui.conversion_runner`` (and its
+``ConversionService`` / ``Job`` machinery) in an async-friendly interface that
+can push real-time progress and log updates back to the Flet event loop without
+blocking the UI thread.
+
+Key design decisions
+--------------------
+* All heavy work is offloaded to daemon threads. The Flet page event loop
+  is never blocked.
+* Progress and log callbacks are scheduled back onto the Flet page via
+  ``page.run_task()`` so Flet's session isolation remains intact.
+* Cancellation is cooperative: the underlying job's ``cancel_requested``
+  flag is set, and the runner checks it at chunk boundaries.
+* The module is a pure adapter – it does NOT duplicate any processing logic
+  from the core pipeline.
+"""
+
+from __future__ import annotations
+
+import asyncio
+import os
+import tempfile
+import threading
+import time
+import traceback
+from pathlib import Path
+from typing import Any, Callable, Dict, List, Optional
+
+import flet as ft
+
+from abogen.utils import (
+    get_gpu_acceleration,
+    get_user_cache_path,
+    get_user_output_path,
+    load_numpy_kpipeline,
+    prevent_sleep_end,
+    prevent_sleep_start,
+)
+from abogen.webui.service import (
+    ConversionService,
+    Job,
+    JobStatus,
+    PendingJob,
+    build_service,
+)
+from abogen.webui.conversion_runner import run_conversion_job
+
+from ..state import AppState
+
+# ---------------------------------------------------------------------------
+# Module-level singleton ConversionService (shared across sessions, as in the
+# web UI – but each job carries its own output folder keyed by session).
+# ---------------------------------------------------------------------------
+
+_SERVICE_LOCK = threading.Lock()
+_SERVICE: Optional[ConversionService] = None
+
+
+def _get_service() -> ConversionService:
+    """
+    Return (creating if necessary) the module-level ConversionService.
+
+    The service manages the background worker thread and persistent job state.
+    Thread-safe via a module-level lock.
+    """
+    global _SERVICE
+    with _SERVICE_LOCK:
+        if _SERVICE is None:
+            output_root = Path(get_user_output_path("frontend"))
+            uploads_root = Path(get_user_cache_path("frontend/uploads"))
+            _SERVICE = build_service(
+                runner=run_conversion_job,
+                output_root=output_root,
+                uploads_root=uploads_root,
+            )
+    return _SERVICE
+
+
+# ---------------------------------------------------------------------------
+# Public conversion bridge
+# ---------------------------------------------------------------------------
+
+
+class ConversionBridge:
+    """
+    Thin adapter between the Flet UI session and the core conversion pipeline.
+
+    One ``ConversionBridge`` instance is created per Flet page (session) and
+    is responsible for:
+    1. Accepting a conversion request from the UI.
+    2. Writing the input text to a temp file if needed.
+    3. Submitting the job to ``ConversionService``.
+    4. Polling the job from a daemon thread and forwarding progress/logs to
+       the Flet page via ``page.run_task()``.
+    5. Providing a ``cancel()`` method that sets the cooperative flag.
+    """
+
+    def __init__(self, page: ft.Page, state: AppState) -> None:
+        """
+        Initialise the bridge.
+
+        Args:
+            page: The Flet ``Page`` for this session. Used to schedule
+                  UI callbacks on the correct event loop.
+            state: The session's ``AppState`` instance.
+        """
+        self._page = page
+        self._state = state
+        self._current_job: Optional[Job] = None
+        self._poll_thread: Optional[threading.Thread] = None
+        self._stop_poll = threading.Event()
+        self._seen_log_count = 0
+
+    # ------------------------------------------------------------------
+    # Public API
+    # ------------------------------------------------------------------
+
+    def start(
+        self,
+        *,
+        input_file: str,
+        voice: str,
+        lang_code: str,
+        speed: float,
+        output_format: str,
+        subtitle_mode: str,
+        subtitle_format: str,
+        use_gpu: bool,
+        save_option: str,
+        output_folder: Optional[str],
+        replace_single_newlines: bool,
+        char_count: int,
+        chapters: Optional[List[Dict[str, Any]]] = None,
+        save_chapters_separately: bool = False,
+        merge_chapters_at_end: bool = True,
+        separate_chapters_format: str = "wav",
+        silence_between_chapters: float = 2.0,
+        max_subtitle_words: int = 50,
+        chapter_intro_delay: float = 0.5,
+        read_title_intro: bool = False,
+        read_closing_outro: bool = True,
+        auto_prefix_chapter_titles: bool = True,
+        normalize_chapter_opening_caps: bool = True,
+        tts_provider: str = "kokoro",
+        supertonic_total_steps: int = 5,
+        chunk_level: str = "paragraph",
+        generate_epub3: bool = False,
+        word_substitutions_enabled: bool = False,
+        word_substitutions_list: str = "",
+        case_sensitive_substitutions: bool = False,
+        replace_all_caps: bool = False,
+        replace_numerals: bool = False,
+        fix_nonstandard_punctuation: bool = False,
+    ) -> None:
+        """
+        Submit a conversion job and begin the progress-polling loop.
+
+        This method returns immediately; all heavy work runs on daemon threads.
+        UI callbacks (``state.on_log``, ``state.on_progress``,
+        ``state.on_conversion_finished``) are scheduled on the Flet event loop.
+
+        Args:
+            input_file: Absolute path to the text/epub/pdf input file.
+            voice: Kokoro voice formula string.
+            lang_code: Single-char language code.
+            speed: Playback speed multiplier (0.1 – 2.0).
+            output_format: Audio container key (``'wav'``, ``'mp3'``, …).
+            subtitle_mode: Subtitle generation mode string.
+            subtitle_format: Subtitle container key (``'srt'``, ``'ass_wide'``, …).
+            use_gpu: Whether to request GPU acceleration.
+            save_option: Save-location strategy string.
+            output_folder: Explicit output folder or None.
+            replace_single_newlines: Pre-processing flag.
+            char_count: Pre-computed character count for ETR estimation.
+            chapters: Optional list of chapter dicts for epub/pdf.
+            save_chapters_separately: Split chapters into separate files.
+            merge_chapters_at_end: Merge chapter files into one after generation.
+            separate_chapters_format: Format for individual chapter files.
+            silence_between_chapters: Silence gap (seconds) between chapters.
+            max_subtitle_words: Maximum words per subtitle block.
+            chapter_intro_delay: Silence before chapter title announcement (s).
+            read_title_intro: Announce book title at the start.
+            read_closing_outro: Announce book title at the end.
+            auto_prefix_chapter_titles: Prepend "Chapter N." to titles.
+            normalize_chapter_opening_caps: Fix ALL-CAPS opening lines.
+            tts_provider: ``'kokoro'`` or ``'supertonic'``.
+            supertonic_total_steps: Quality steps for the Supertonic pipeline.
+            chunk_level: ``'paragraph'`` or ``'sentence'`` chunking granularity.
+            generate_epub3: Also produce an EPUB3 audiobook package.
+            word_substitutions_enabled: Toggle word-substitution pre-processing.
+            word_substitutions_list: Newline-delimited ``word|replacement`` rules.
+            case_sensitive_substitutions: Case-sensitive matching for substitutions.
+            replace_all_caps: Lowercase ALL-CAPS words.
+            replace_numerals: Convert digits to spoken words.
+            fix_nonstandard_punctuation: Normalise curly quotes etc.
+        """
+        if self._state.is_converting:
+            return
+
+        # Resolve the effective output folder
+        resolved_output: Optional[Path] = self._resolve_output_folder(
+            save_option=save_option,
+            output_folder=output_folder,
+            input_file=input_file,
+        )
+
+        # Store the input file as a Path
+        stored_path = Path(input_file)
+        original_filename = stored_path.name
+
+        # Block signals until the job is submitted
+        prevent_sleep_start()
+        self._state.is_converting = True
+        self._state.is_cancelled = False
+        self._state.progress = 0.0
+        self._state.etr_seconds = None
+        self._state.log_lines = []
+        self._seen_log_count = 0
+
+        # Enqueue the job on the service
+        service = _get_service()
+        job = service.enqueue(
+            original_filename=original_filename,
+            stored_path=stored_path,
+            language=lang_code,
+            voice=voice,
+            speed=speed,
+            tts_provider=tts_provider,
+            supertonic_total_steps=supertonic_total_steps,
+            use_gpu=use_gpu,
+            subtitle_mode=subtitle_mode,
+            output_format=output_format,
+            save_mode=self._save_mode_key(save_option),
+            output_folder=resolved_output,
+            replace_single_newlines=replace_single_newlines,
+            subtitle_format=subtitle_format,
+            total_characters=char_count,
+            chapters=chapters or [],
+            save_chapters_separately=save_chapters_separately,
+            merge_chapters_at_end=merge_chapters_at_end,
+            separate_chapters_format=separate_chapters_format,
+            silence_between_chapters=silence_between_chapters,
+            max_subtitle_words=max_subtitle_words,
+            chapter_intro_delay=chapter_intro_delay,
+            read_title_intro=read_title_intro,
+            read_closing_outro=read_closing_outro,
+            auto_prefix_chapter_titles=auto_prefix_chapter_titles,
+            normalize_chapter_opening_caps=normalize_chapter_opening_caps,
+            chunk_level=chunk_level,
+            generate_epub3=generate_epub3,
+        )
+        self._current_job = job
+
+        # Persist word-substitution settings to config so the runner picks them up
+        self._state.word_substitutions_enabled = word_substitutions_enabled
+        self._state.word_substitutions_list = word_substitutions_list
+        self._state.case_sensitive_substitutions = case_sensitive_substitutions
+        self._state.replace_all_caps = replace_all_caps
+        self._state.replace_numerals = replace_numerals
+        self._state.fix_nonstandard_punctuation = fix_nonstandard_punctuation
+        self._state.persist_config()
+
+        # Start the poll thread
+        self._stop_poll.clear()
+        self._poll_thread = threading.Thread(
+            target=self._poll_job_loop, daemon=True, name="abogen-poll"
+        )
+        self._poll_thread.start()
+
+    def cancel(self) -> None:
+        """
+        Request cancellation of the currently running job.
+
+        Sets the cooperative flag on the underlying ``Job`` object; the runner
+        will stop after completing the current text chunk.
+        """
+        if self._current_job is not None:
+            self._state.is_cancelled = True
+            try:
+                _get_service().cancel(self._current_job.id)
+            except Exception:
+                pass
+
+    # ------------------------------------------------------------------
+    # Internal helpers
+    # ------------------------------------------------------------------
+
+    @staticmethod
+    def _save_mode_key(option: str) -> str:
+        """
+        Convert the human-readable save option to the service's internal key.
+
+        Args:
+            option: UI-facing string (``'Save next to input file'``, …).
+
+        Returns:
+            Service key string.
+        """
+        mapping = {
+            "Save next to input file": "save_next_to_input",
+            "Save to Desktop": "save_to_desktop",
+            "Choose output folder": "custom",
+        }
+        return mapping.get(option, "save_next_to_input")
+
+    @staticmethod
+    def _resolve_output_folder(
+        save_option: str,
+        output_folder: Optional[str],
+        input_file: str,
+    ) -> Optional[Path]:
+        """
+        Return the output ``Path`` based on the save option, or None for
+        the "next to input" strategy (the runner handles that internally).
+
+        Args:
+            save_option: UI-facing save strategy string.
+            output_folder: Explicit path when ``save_option`` is ``'Choose output folder'``.
+            input_file: Path to the source file for the ``'Save to Desktop'`` strategy.
+
+        Returns:
+            Resolved ``Path`` or ``None``.
+        """
+        if save_option == "Choose output folder" and output_folder:
+            p = Path(output_folder)
+            p.mkdir(parents=True, exist_ok=True)
+            return p
+        if save_option == "Save to Desktop":
+            desktop = Path.home() / "Desktop"
+            desktop.mkdir(exist_ok=True)
+            return desktop
+        # "Save next to input file" – let the runner decide
+        return None
+
+    def _poll_job_loop(self) -> None:
+        """
+        Background daemon loop that polls the current Job for updates.
+
+        Runs until the job enters a terminal state or until ``_stop_poll``
+        is set.  Uses ``page.run_task()`` to schedule UI updates on the Flet
+        event loop without triggering thread-safety violations.
+        """
+        job = self._current_job
+        if job is None:
+            return
+
+        service = _get_service()
+        POLL_INTERVAL = 0.25  # seconds
+
+        while not self._stop_poll.is_set():
+            # Re-fetch the current job state (it's mutated in-place by the runner)
+            current = service.get_job(job.id)
+            if current is None:
+                break
+
+            # Forward new log lines
+            new_logs = current.logs[self._seen_log_count:]
+            self._seen_log_count += len(new_logs)
+            for log_entry in new_logs:
+                level = getattr(log_entry, "level", "info")
+                message = getattr(log_entry, "message", str(log_entry))
+                self._schedule_log(message, level)
+
+            # Forward progress
+            if current.progress is not None:
+                etr = getattr(current, "estimated_time_remaining", None)
+                self._schedule_progress(float(current.progress), etr)
+
+            # Check for terminal states
+            status = current.status
+            if status in (
+                JobStatus.COMPLETED,
+                JobStatus.FAILED,
+                JobStatus.CANCELLED,
+            ):
+                output_path: Optional[str] = None
+                if current.result and current.result.audio_path:
+                    output_path = str(current.result.audio_path)
+                if status == JobStatus.COMPLETED:
+                    finish_msg = "Conversion completed successfully."
+                elif status == JobStatus.CANCELLED:
+                    finish_msg = "Cancelled"
+                else:
+                    finish_msg = f"Conversion failed: {current.error or 'Unknown error'}"
+
+                self._schedule_finished(finish_msg, output_path)
+                break
+
+            time.sleep(POLL_INTERVAL)
+
+        prevent_sleep_end()
+        self._state.is_converting = False
+
+    def _schedule_log(self, message: str, level: str) -> None:
+        """Schedule a log update on the Flet event loop."""
+        state = self._state
+        page = self._page
+        state.append_log(message, level)
+
+        async def _update() -> None:
+            cb = state.on_log
+            if cb:
+                cb(message, level)
+            try:
+                page.update()
+            except Exception:
+                pass
+
+        try:
+            page.run_task(_update)
+        except Exception:
+            pass
+
+    def _schedule_progress(self, fraction: float, etr: Optional[float]) -> None:
+        """Schedule a progress update on the Flet event loop."""
+        state = self._state
+        page = self._page
+        state.progress = max(0.0, min(1.0, fraction))
+        state.etr_seconds = etr
+
+        async def _update() -> None:
+            cb = state.on_progress
+            if cb:
+                cb(fraction, etr)
+            try:
+                page.update()
+            except Exception:
+                pass
+
+        try:
+            page.run_task(_update)
+        except Exception:
+            pass
+
+    def _schedule_finished(
+        self, message: str, output_path: Optional[str]
+    ) -> None:
+        """Schedule a completion notification on the Flet event loop."""
+        state = self._state
+        page = self._page
+        state.last_output_path = output_path
+        self._stop_poll.set()
+
+        async def _update() -> None:
+            state.is_converting = False
+            state.progress = 1.0
+            state.last_output_path = output_path
+            cb = state.on_conversion_finished
+            if cb:
+                cb(message, output_path)
+            try:
+                page.update()
+            except Exception:
+                pass
+
+        try:
+            page.run_task(_update)
+        except Exception:
+            pass

abogen/frontend/utils/helpers.py

@@ -0,0 +1,313 @@
+"""
+Frontend-specific utilities for the Abogen Flet application.
+
+Contains helpers for:
+- Human-readable size / duration formatting
+- Voice formula parsing and display
+- File-type detection
+- ETR (Estimated Time Remaining) formatting
+- Path resolution that adapts to desktop vs. web context
+"""
+
+from __future__ import annotations
+
+import os
+import re
+from pathlib import Path
+from typing import List, Optional, Tuple
+
+from abogen.constants import (
+    LANGUAGE_DESCRIPTIONS,
+    SUPPORTED_INPUT_FORMATS,
+    SUPPORTED_SOUND_FORMATS,
+    SUBTITLE_FORMATS,
+    VOICES_INTERNAL,
+)
+
+
+# ---------------------------------------------------------------------------
+# Size / duration helpers
+# ---------------------------------------------------------------------------
+
+
+def human_readable_size(size_bytes: int, decimal_places: int = 2) -> str:
+    """
+    Convert a byte count into a human-readable string.
+
+    Args:
+        size_bytes: Number of bytes.
+        decimal_places: Significant decimal digits in the output.
+
+    Returns:
+        A string like ``"3.14 MB"`` or ``"1.00 KB"``.
+    """
+    for unit in ("B", "KB", "MB", "GB", "TB"):
+        if size_bytes < 1024.0:
+            return f"{size_bytes:.{decimal_places}f} {unit}"
+        size_bytes /= 1024.0  # type: ignore[assignment]
+    return f"{size_bytes:.{decimal_places}f} PB"
+
+
+def format_duration(seconds: float) -> str:
+    """
+    Format a duration in seconds as ``HH:MM:SS``.
+
+    Args:
+        seconds: Non-negative floating-point duration.
+
+    Returns:
+        A colon-delimited time string, e.g. ``"00:03:42"``.
+    """
+    total = max(0, int(seconds))
+    h, remainder = divmod(total, 3600)
+    m, s = divmod(remainder, 60)
+    return f"{h:02d}:{m:02d}:{s:02d}"
+
+
+def format_etr(etr_seconds: Optional[float]) -> str:
+    """
+    Format an estimated time remaining value for the UI.
+
+    Args:
+        etr_seconds: Seconds remaining, or None when unknown.
+
+    Returns:
+        Human-readable string such as ``"~3 min 42 sec"`` or ``"Calculating…"``.
+    """
+    if etr_seconds is None:
+        return "Calculating…"
+    total = max(0, int(etr_seconds))
+    if total < 60:
+        return f"~{total} sec"
+    m, s = divmod(total, 60)
+    if m < 60:
+        return f"~{m} min {s} sec"
+    h, m = divmod(m, 60)
+    return f"~{h} h {m} min"
+
+
+# ---------------------------------------------------------------------------
+# File helpers
+# ---------------------------------------------------------------------------
+
+
+SUPPORTED_EXTENSIONS: Tuple[str, ...] = (
+    ".txt",
+    ".epub",
+    ".pdf",
+    ".md",
+    ".markdown",
+    ".srt",
+    ".ass",
+    ".vtt",
+)
+"""All file extensions that the drop-zone accepts."""
+
+
+def detect_file_type(file_path: str) -> str:
+    """
+    Return a normalised file-type token for the given path.
+
+    Args:
+        file_path: Absolute or relative path to the input file.
+
+    Returns:
+        One of ``'txt'``, ``'epub'``, ``'pdf'``, ``'markdown'``,
+        ``'subtitle'``, or ``'unknown'``.
+    """
+    ext = Path(file_path).suffix.lower()
+    if ext == ".epub":
+        return "epub"
+    if ext == ".pdf":
+        return "pdf"
+    if ext in (".md", ".markdown"):
+        return "markdown"
+    if ext in (".srt", ".ass", ".vtt"):
+        return "subtitle"
+    if ext == ".txt":
+        return "txt"
+    return "unknown"
+
+
+def is_supported_file(file_path: str) -> bool:
+    """
+    Return True when the file extension is in the supported set.
+
+    Args:
+        file_path: Path whose extension is inspected.
+    """
+    return Path(file_path).suffix.lower() in SUPPORTED_EXTENSIONS
+
+
+def is_book_type(file_type: str) -> bool:
+    """
+    Return True for file types that contain chapters / pages.
+
+    Args:
+        file_type: Token from ``detect_file_type()``.
+    """
+    return file_type in ("epub", "pdf", "markdown")
+
+
+# ---------------------------------------------------------------------------
+# Voice helpers
+# ---------------------------------------------------------------------------
+
+
+def voice_lang_code(voice: str) -> str:
+    """
+    Extract the language code character from a Kokoro voice name.
+
+    The first character of every internal voice name encodes the language
+    (e.g. ``'a'`` for American English, ``'b'`` for British English).
+
+    Args:
+        voice: Raw voice string like ``'af_heart'`` or a formula.
+
+    Returns:
+        Single lowercase character, defaulting to ``'a'`` on failure.
+    """
+    if not voice:
+        return "a"
+    # For plain voice IDs the first char is the language
+    if voice[0].isalpha() and "_" in voice[:4]:
+        return voice[0].lower()
+    # Formula: extract first alpha char
+    match = re.search(r"\b([a-z])", voice)
+    return match.group(1) if match else "a"
+
+
+def language_label(lang_code: str) -> str:
+    """
+    Return the human-readable label for a language code.
+
+    Args:
+        lang_code: Single-character code (``'a'``, ``'b'``, …).
+
+    Returns:
+        Display string, e.g. ``"American English"``.
+    """
+    return LANGUAGE_DESCRIPTIONS.get(lang_code, lang_code.upper())
+
+
+def grouped_voices() -> List[Tuple[str, List[str]]]:
+    """
+    Return the internal voice list grouped by language for display.
+
+    Returns:
+        List of ``(language_label, [voice_id, …])`` tuples.
+    """
+    groups: dict[str, List[str]] = {}
+    for v in VOICES_INTERNAL:
+        lang = language_label(v[0])
+        groups.setdefault(lang, []).append(v)
+    return sorted(groups.items())
+
+
+def voice_display_name(voice_id: str) -> str:
+    """
+    Convert a raw voice ID like ``'af_heart'`` to a prettier display name.
+
+    Args:
+        voice_id: Raw internal voice identifier.
+
+    Returns:
+        Formatted string, e.g. ``"af_heart"`` (unchanged; may be enhanced later).
+    """
+    return voice_id
+
+
+def parse_voice_formula(formula: str) -> List[Tuple[str, float]]:
+    """
+    Parse a Kokoro voice mix formula into a list of ``(voice_id, weight)`` tuples.
+
+    Example:
+        ``"af_heart*0.7+am_adam*0.3"`` → ``[('af_heart', 0.7), ('am_adam', 0.3)]``
+
+    Args:
+        formula: Space- or ``+``-joined mix formula string.
+
+    Returns:
+        Parsed list; empty if parsing fails.
+    """
+    parts: List[Tuple[str, float]] = []
+    for token in re.split(r"[+\s]+", formula.strip()):
+        token = token.strip()
+        if not token:
+            continue
+        if "*" in token:
+            name, _, weight_str = token.partition("*")
+            try:
+                parts.append((name.strip(), float(weight_str.strip())))
+            except ValueError:
+                pass
+        else:
+            # Bare voice id — assume full weight
+            if token in VOICES_INTERNAL:
+                parts.append((token, 1.0))
+    return parts
+
+
+# ---------------------------------------------------------------------------
+# Number formatting
+# ---------------------------------------------------------------------------
+
+
+def format_number(n: int) -> str:
+    """
+    Format an integer with thousands separators.
+
+    Args:
+        n: Integer value.
+
+    Returns:
+        Formatted string, e.g. ``"1,234,567"``.
+    """
+    return f"{n:,}"
+
+
+# ---------------------------------------------------------------------------
+# Path helpers
+# ---------------------------------------------------------------------------
+
+
+def safe_basename(path: Optional[str]) -> str:
+    """
+    Return the basename of a path, or an empty string when path is None/empty.
+
+    Args:
+        path: Optional file-system path.
+    """
+    if not path:
+        return ""
+    return os.path.basename(path)
+
+
+def output_format_label(fmt: str) -> str:
+    """
+    Return a display label for an audio output format key.
+
+    Args:
+        fmt: Lowercase format key (``'wav'``, ``'mp3'``, …).
+    """
+    labels = {
+        "wav": "WAV (lossless)",
+        "flac": "FLAC (lossless compressed)",
+        "mp3": "MP3",
+        "opus": "Opus (best compression)",
+        "m4b": "M4B (with chapters)",
+    }
+    return labels.get(fmt, fmt.upper())
+
+
+def subtitle_format_label(key: str) -> str:
+    """
+    Return the display label for a subtitle format key.
+
+    Args:
+        key: Internal subtitle format key (e.g. ``'ass_centered_narrow'``).
+    """
+    for k, label in SUBTITLE_FORMATS:
+        if k == key:
+            return label
+    return key

abogen/frontend/utils/theme.py

@@ -0,0 +1,264 @@
+"""
+Design tokens and theme configuration for the Abogen Flet frontend.
+
+This module defines the application's complete colour palette, typography
+scale, spacing constants, and border radii in one canonical place.
+All component modules import from here; changing a value here propagates
+instantly across the entire UI.
+
+Flet's ``ft.Theme`` uses ``ColorScheme``, but for custom widgets we paint
+directly with hex colours drawn from ``LIGHT`` and ``DARK`` palettes.
+"""
+
+from __future__ import annotations
+
+import flet as ft
+from dataclasses import dataclass
+
+
+# ---------------------------------------------------------------------------
+# Colour palettes
+# ---------------------------------------------------------------------------
+
+
+@dataclass(frozen=True)
+class _Palette:
+    """A complete colour palette for one theme mode."""
+
+    # Backgrounds
+    bg_base: str        # Deepest background (window / page)
+    bg_surface: str     # Cards, panels, dialogs
+    bg_elevated: str    # Slightly raised elements (toolbar, sidebar)
+    bg_input: str       # Text-field / dropdown backgrounds
+
+    # Brand accent
+    accent: str         # Primary interactive colour (buttons, links)
+    accent_muted: str   # Hover tint over accents
+    accent_on: str      # Text drawn on top of accent fills
+
+    # Semantic
+    success: str
+    error: str
+    warning: str
+    info: str
+
+    # Text hierarchy
+    text_primary: str
+    text_secondary: str
+    text_disabled: str
+    text_on_accent: str
+
+    # Borders / dividers
+    border: str
+    border_focused: str
+    divider: str
+
+    # Specific UI atoms
+    drop_zone_border: str
+    drop_zone_bg: str
+    drop_zone_active_border: str
+    drop_zone_active_bg: str
+    log_bg: str
+    log_text: str
+    progress_bar_bg: str
+    progress_bar_fill: str
+    sidebar_bg: str
+    sidebar_selected_bg: str
+    sidebar_selected_text: str
+    nav_indicator: str
+
+
+DARK = _Palette(
+    bg_base="#0f1117",
+    bg_surface="#181b23",
+    bg_elevated="#1e2230",
+    bg_input="#252a38",
+
+    accent="#5b8af5",
+    accent_muted="#3a5fc4",
+    accent_on="#ffffff",
+
+    success="#42ad4a",
+    error="#e84e3c",
+    warning="#f5a623",
+    info="#5b8af5",
+
+    text_primary="#e8eaf0",
+    text_secondary="#9ba3b8",
+    text_disabled="#4e5568",
+    text_on_accent="#ffffff",
+
+    border="#2c3147",
+    border_focused="#5b8af5",
+    divider="#252a38",
+
+    drop_zone_border="#3a4466",
+    drop_zone_bg="#151928",
+    drop_zone_active_border="#42ad4a",
+    drop_zone_active_bg="#0d1f10",
+    log_bg="#0d1117",
+    log_text="#b0b8cc",
+    progress_bar_bg="#1e2230",
+    progress_bar_fill="#5b8af5",
+    sidebar_bg="#13161f",
+    sidebar_selected_bg="#252a38",
+    sidebar_selected_text="#5b8af5",
+    nav_indicator="#5b8af5",
+)
+
+LIGHT = _Palette(
+    bg_base="#f4f5f8",
+    bg_surface="#ffffff",
+    bg_elevated="#edf0f5",
+    bg_input="#f0f2f7",
+
+    accent="#3a5fc4",
+    accent_muted="#2a4fae",
+    accent_on="#ffffff",
+
+    success="#2e9437",
+    error="#c0392b",
+    warning="#d4870a",
+    info="#3a5fc4",
+
+    text_primary="#1a1d27",
+    text_secondary="#5a6172",
+    text_disabled="#9ba3b8",
+    text_on_accent="#ffffff",
+
+    border="#dce0ea",
+    border_focused="#3a5fc4",
+    divider="#e8ebf2",
+
+    drop_zone_border="#a8b4d0",
+    drop_zone_bg="#f7f8fd",
+    drop_zone_active_border="#2e9437",
+    drop_zone_active_bg="#f0fff1",
+    log_bg="#f8f9fc",
+    log_text="#3d4358",
+    progress_bar_bg="#e4e8f0",
+    progress_bar_fill="#3a5fc4",
+    sidebar_bg="#eff1f5",
+    sidebar_selected_bg="#dde3f2",
+    sidebar_selected_text="#3a5fc4",
+    nav_indicator="#3a5fc4",
+)
+
+
+# ---------------------------------------------------------------------------
+# Typography
+# ---------------------------------------------------------------------------
+
+FONT_FAMILY = "Inter, Segoe UI, Roboto, system-ui, sans-serif"
+FONT_SIZE_XS = 11
+FONT_SIZE_SM = 12
+FONT_SIZE_BASE = 14
+FONT_SIZE_MD = 16
+FONT_SIZE_LG = 20
+FONT_SIZE_XL = 26
+FONT_SIZE_DISPLAY = 34
+
+
+# ---------------------------------------------------------------------------
+# Spacing scale (pixels)
+# ---------------------------------------------------------------------------
+
+SPACE_XS = 4
+SPACE_SM = 8
+SPACE_MD = 12
+SPACE_LG = 16
+SPACE_XL = 24
+SPACE_2XL = 32
+SPACE_3XL = 48
+
+
+# ---------------------------------------------------------------------------
+# Border radii
+# ---------------------------------------------------------------------------
+
+RADIUS_SM = 6
+RADIUS_MD = 10
+RADIUS_LG = 16
+RADIUS_FULL = 999  # Pill-shaped
+
+
+# ---------------------------------------------------------------------------
+# Flet ColorScheme builders
+# ---------------------------------------------------------------------------
+
+
+def build_color_scheme(palette: _Palette) -> ft.ColorScheme:
+    """
+    Construct a ``ft.ColorScheme`` from a ``_Palette`` object.
+
+    Args:
+        palette: The ``DARK`` or ``LIGHT`` palette.
+
+    Returns:
+        A fully-populated Flet ``ColorScheme``.
+    """
+    return ft.ColorScheme(
+        primary=palette.accent,
+        on_primary=palette.accent_on,
+        primary_container=palette.accent_muted,
+        secondary=palette.accent,
+        on_secondary=palette.text_on_accent,
+        surface=palette.bg_surface,
+        on_surface=palette.text_primary,
+        on_surface_variant=palette.text_secondary,
+        error=palette.error,
+        on_error=palette.text_on_accent,
+        outline=palette.border,
+    )
+
+
+def build_text_theme() -> ft.TextTheme:
+    """
+    Construct a ``ft.TextTheme`` using the application's type scale.
+
+    Returns:
+        A Flet ``TextTheme`` with consistent font-size assignments.
+    """
+    return ft.TextTheme(
+        display_large=ft.TextStyle(size=FONT_SIZE_DISPLAY, weight=ft.FontWeight.W_700),
+        headline_large=ft.TextStyle(size=FONT_SIZE_XL, weight=ft.FontWeight.W_700),
+        headline_medium=ft.TextStyle(size=FONT_SIZE_LG, weight=ft.FontWeight.W_600),
+        title_large=ft.TextStyle(size=FONT_SIZE_MD, weight=ft.FontWeight.W_600),
+        title_medium=ft.TextStyle(size=FONT_SIZE_BASE, weight=ft.FontWeight.W_500),
+        body_large=ft.TextStyle(size=FONT_SIZE_BASE),
+        body_medium=ft.TextStyle(size=FONT_SIZE_SM),
+        label_large=ft.TextStyle(size=FONT_SIZE_SM, weight=ft.FontWeight.W_500),
+        label_medium=ft.TextStyle(size=FONT_SIZE_XS),
+    )
+
+
+def make_theme(dark: bool) -> ft.Theme:
+    """
+    Build a complete Flet ``Theme`` for the requested mode.
+
+    Args:
+        dark: True for dark-mode theme, False for light-mode theme.
+
+    Returns:
+        A configured ``ft.Theme`` instance.
+    """
+    palette = DARK if dark else LIGHT
+    return ft.Theme(
+        color_scheme=build_color_scheme(palette),
+        text_theme=build_text_theme(),
+        color_scheme_seed=palette.accent,
+        use_material3=True,
+    )
+
+
+def get_palette(page: ft.Page) -> _Palette:
+    """
+    Return the active colour palette for the given page.
+
+    Args:
+        page: The Flet ``Page`` instance.
+
+    Returns:
+        ``DARK`` or ``LIGHT`` depending on the page's theme mode.
+    """
+    return DARK if page.theme_mode == ft.ThemeMode.DARK else LIGHT

abogen/frontend/views/__init__.py

@@ -0,0 +1,6 @@
+"""Views sub-package for the Abogen Flet frontend."""
+from .dashboard import DashboardView
+from .settings import SettingsView
+from .queue_view import QueueView
+
+__all__ = ["DashboardView", "SettingsView", "QueueView"]

abogen/frontend/views/dashboard.py

@@ -0,0 +1,587 @@
+"""
+Dashboard view – the primary conversion screen.
+
+Hosts the file drop-zone, voice/speed/format controls, real-time log
+terminal, progress bar, and the Start/Cancel/Finish action row.
+
+All heavy work is delegated to ConversionBridge which runs on daemon
+threads and schedules UI updates back onto the Flet event loop.
+"""
+from __future__ import annotations
+
+import os
+import tempfile
+from pathlib import Path
+from typing import Optional
+
+import flet as ft
+
+from ..state import AppState
+from ..utils.helpers import (
+    detect_file_type, human_readable_size, format_number,
+    format_etr, grouped_voices, output_format_label,
+    subtitle_format_label, is_book_type, voice_lang_code, SUPPORTED_EXTENSIONS
+)
+from ..utils.theme import get_palette, RADIUS_MD, RADIUS_SM, SPACE_SM, SPACE_MD, SPACE_LG, SPACE_XL
+from ..utils.conversion_bridge import ConversionBridge
+from ..components import (
+    build_drop_zone, build_log_terminal, log_entry,
+    build_primary_button, build_secondary_button,
+    build_card, build_section_header, labelled_row, show_snack,
+)
+from abogen.constants import (
+    SUBTITLE_FORMATS, SUPPORTED_LANGUAGES_FOR_SUBTITLE_GENERATION,
+    LANGUAGE_DESCRIPTIONS, VOICES_INTERNAL,
+)
+from abogen.utils import get_gpu_acceleration, get_user_cache_path, calculate_text_length, clean_text
+
+
+class DashboardView:
+    """
+    The main conversion dashboard.
+
+    Instantiated once per Flet session and mounted as a ``ft.Column``
+    inside the page's content area.
+    """
+
+    def __init__(self, page: ft.Page, state: AppState) -> None:
+        self._page = page
+        self._state = state
+        self._bridge = ConversionBridge(page, state)
+
+        # Internal refs
+        self._log_list: Optional[ft.ListView] = None
+        self._progress_bar: Optional[ft.ProgressBar] = None
+        self._etr_label: Optional[ft.Text] = None
+        self._drop_zone_ref: Optional[ft.GestureDetector] = None
+        self._drop_zone_container: Optional[ft.Container] = None
+        self._file_picker: Optional[ft.FilePicker] = None
+
+        # Wire state callbacks
+        state.on_log = self._on_log
+        state.on_progress = self._on_progress
+        state.on_conversion_finished = self._on_finished
+
+        # Build UI refs
+        self._voice_dd: Optional[ft.Dropdown] = None
+        self._speed_slider: Optional[ft.Slider] = None
+        self._speed_label: Optional[ft.Text] = None
+        self._format_dd: Optional[ft.Dropdown] = None
+        self._subtitle_dd: Optional[ft.Dropdown] = None
+        self._subtitle_fmt_dd: Optional[ft.Dropdown] = None
+        self._gpu_switch: Optional[ft.Switch] = None
+        self._start_btn: Optional[ft.ElevatedButton] = None
+        self._cancel_btn: Optional[ft.OutlinedButton] = None
+        self._finish_col: Optional[ft.Column] = None
+        self._controls_col: Optional[ft.Column] = None
+        self._log_section: Optional[ft.Container] = None
+        self._progress_col: Optional[ft.Column] = None
+
+    # ------------------------------------------------------------------
+    # Build
+    # ------------------------------------------------------------------
+
+    def build(self) -> ft.Column:
+        """Return the complete dashboard column."""
+        p = self._page
+        dark = p.theme_mode == ft.ThemeMode.DARK
+        pal = get_palette(p)
+        if self._file_picker is None:
+            self._file_picker = ft.FilePicker()
+
+        # --- Drop zone ---
+        self._drop_zone_container = ft.Container()
+        self._refresh_drop_zone()
+
+        # --- Voice selector ---
+        voice_items = []
+        for lang_label, voices in grouped_voices():
+            voice_items.append(ft.dropdown.Option(key=f"__hdr_{lang_label}", text=f"── {lang_label} ──", disabled=True))
+            for v in voices:
+                voice_items.append(ft.dropdown.Option(key=v, text=v))
+
+        self._voice_dd = ft.Dropdown(
+            options=voice_items,
+            value=self._state.selected_voice,
+            on_select=self._on_voice_changed,
+            dense=True,
+            expand=True,
+            border_radius=RADIUS_SM,
+        )
+
+        # --- Speed slider ---
+        self._speed_label = ft.Text(f"{self._state.speed:.2f}", size=13, width=40)
+        self._speed_slider = ft.Slider(
+            min=0.1, max=2.0, value=self._state.speed,
+            divisions=190, label="{value}",
+            on_change=self._on_speed_changed,
+            expand=True,
+        )
+
+        # --- Format ---
+        self._format_dd = ft.Dropdown(
+            options=[ft.dropdown.Option(key=k, text=output_format_label(k))
+                     for k in ("wav", "flac", "mp3", "opus", "m4b")],
+            value=self._state.selected_format,
+            on_select=lambda e: self._set_field("selected_format", e.control.value),
+            dense=True, expand=True, border_radius=RADIUS_SM,
+        )
+
+        # --- Subtitle mode ---
+        sub_modes = ["Disabled", "Line", "Sentence", "Sentence + Comma",
+                     "Sentence + Highlighting"] + [f"{i} word{'s' if i > 1 else ''}" for i in range(1, 11)]
+        self._subtitle_dd = ft.Dropdown(
+            options=[ft.dropdown.Option(m) for m in sub_modes],
+            value=self._state.subtitle_mode,
+            on_select=lambda e: self._set_field("subtitle_mode", e.control.value),
+            dense=True, expand=True, border_radius=RADIUS_SM,
+        )
+
+        # --- Subtitle format ---
+        self._subtitle_fmt_dd = ft.Dropdown(
+            options=[ft.dropdown.Option(key=k, text=lbl) for k, lbl in SUBTITLE_FORMATS],
+            value=self._state.subtitle_format,
+            on_select=lambda e: self._set_field("subtitle_format", e.control.value),
+            dense=True, expand=True, border_radius=RADIUS_SM,
+        )
+
+        # --- GPU ---
+        self._gpu_switch = ft.Switch(
+            value=self._state.use_gpu, label="",
+            on_change=lambda e: self._set_field("use_gpu", e.control.value),
+            active_color="#5b8af5" if dark else "#3a5fc4",
+        )
+
+        # --- Log ---
+        log_lv = ft.ListView(expand=True, auto_scroll=True, spacing=1, padding=ft.Padding.all(8))
+        self._log_list = log_lv
+        bg_log = "#0d1117" if dark else "#f8f9fc"
+        bd_log = "#252a38" if dark else "#dce0ea"
+        self._log_section = ft.Container(
+            content=log_lv, bgcolor=bg_log,
+            border=ft.Border.all(1, bd_log),
+            border_radius=RADIUS_SM, height=220,
+            clip_behavior=ft.ClipBehavior.HARD_EDGE,
+            visible=False,
+        )
+
+        # --- Progress ---
+        fill = "#5b8af5" if dark else "#3a5fc4"
+        bg_p = "#1e2230" if dark else "#e4e8f0"
+        self._progress_bar = ft.ProgressBar(
+            value=0, color=fill, bgcolor=bg_p, height=8,
+            border_radius=ft.BorderRadius.all(4), expand=True,
+        )
+        self._etr_label = ft.Text("", size=11, color=pal.text_secondary, text_align=ft.TextAlign.CENTER)
+        self._progress_col = ft.Column([
+            ft.Row([self._progress_bar], spacing=0),
+            self._etr_label,
+        ], spacing=SPACE_SM, horizontal_alignment=ft.CrossAxisAlignment.CENTER, visible=False)
+
+        # --- Buttons ---
+        self._start_btn = build_primary_button(
+            "Start Conversion",
+            icon="play_arrow",
+            on_click=self._on_start,
+            page=p,
+        )
+        self._cancel_btn = build_secondary_button(
+            "Cancel", icon="stop",
+            on_click=self._on_cancel, page=p,
+        )
+        self._cancel_btn.visible = False
+
+        # --- Finish row ---
+        self._finish_col = ft.Column([
+            ft.Row([
+                build_secondary_button("Open File", icon="open_in_new",
+                                       on_click=self._on_open_file, page=p),
+                build_secondary_button("Go to Folder", icon="folder_open",
+                                       on_click=self._on_go_folder, page=p),
+                build_secondary_button("New Conversion", icon="refresh",
+                                       on_click=self._on_reset, page=p),
+            ], wrap=True, spacing=SPACE_SM, run_spacing=SPACE_SM),
+        ], visible=False)
+
+        # --- Controls column ---
+        self._controls_col = ft.Column([
+            build_section_header("Voice & Speed", icon="record_voice_over", page=p),
+            labelled_row("Voice", self._voice_dd, page=p),
+            labelled_row("Speed", ft.Row([self._speed_slider, self._speed_label], expand=True, spacing=SPACE_SM), page=p),
+            ft.Divider(height=1, color=pal.divider),
+            build_section_header("Output", icon="audio_file", page=p),
+            labelled_row("Format", self._format_dd, page=p),
+            labelled_row("Subtitles", self._subtitle_dd, page=p),
+            labelled_row("Subtitle Format", self._subtitle_fmt_dd, page=p),
+            ft.Divider(height=1, color=pal.divider),
+            build_section_header("Processing", icon="memory", page=p),
+            labelled_row("GPU Acceleration", self._gpu_switch, page=p),
+        ], spacing=SPACE_MD)
+
+        outer = ft.Column([
+            self._drop_zone_container,
+            ft.Container(height=SPACE_MD),
+            build_card(self._controls_col, page=p),
+            ft.Container(height=SPACE_SM),
+            self._log_section,
+            self._progress_col,
+            ft.Row([self._start_btn, self._cancel_btn], spacing=SPACE_SM, wrap=True),
+            self._finish_col,
+        ], spacing=SPACE_MD, expand=True, scroll=ft.ScrollMode.AUTO)
+
+        return outer
+
+    # ------------------------------------------------------------------
+    # Drop-zone management
+    # ------------------------------------------------------------------
+
+    def _refresh_drop_zone(self, *, accent: bool = False, error: bool = False, err_msg: str = "") -> None:
+        """Rebuild the drop-zone widget and update its container."""
+        p = self._page
+        s = self._state
+        fname = None; fsize = None; fchars = None
+        if s.selected_file and os.path.exists(s.selected_file):
+            disp = s.displayed_file_path or s.selected_file
+            fname = os.path.basename(disp)
+            try:
+                fsize = human_readable_size(os.path.getsize(s.selected_file))
+            except Exception:
+                fsize = ""
+            if s.char_count:
+                fchars = format_number(s.char_count)
+
+        label = err_msg if error else "Drag & drop your file here or click to browse"
+        sub = "Supports .txt · .epub · .pdf · .md · .srt · .ass · .vtt"
+
+        dz = build_drop_zone(
+            on_pick=self._open_file_picker,
+            label=label, sub_label=sub,
+            accent=accent, error=error,
+            filename=fname, file_size=fsize, char_count=fchars,
+            page=p,
+        )
+        if self._drop_zone_container is not None:
+            self._drop_zone_container.content = dz
+        self._drop_zone_ref = dz
+
+    # ------------------------------------------------------------------
+    # File picking
+    # ------------------------------------------------------------------
+
+    def _open_file_picker(self) -> None:
+        """Open the native file picker dialog."""
+        self._page.run_task(self._pick_files_async)
+
+    async def _pick_files_async(self) -> None:
+        """Run the file picker using Flet's async service API."""
+        picker = self._file_picker
+        if picker is None:
+            picker = ft.FilePicker()
+            self._file_picker = picker
+
+        try:
+            files = await picker.pick_files(
+                dialog_title="Select Input File",
+                file_type=ft.FilePickerFileType.CUSTOM,
+                allowed_extensions=["txt", "epub", "pdf", "md", "markdown", "srt", "ass", "vtt"],
+                allow_multiple=False,
+            )
+        except Exception as ex:
+            self._refresh_drop_zone(error=True, err_msg="Could not open file picker.")
+            show_snack(self._page, f"File picker error: {ex}", error=True)
+            self._page.update()
+            return
+        if not files:
+            return
+        file_path = files[0].path
+        if not file_path or not os.path.exists(file_path):
+            return
+        self._load_file(file_path)
+
+    def _load_file(self, file_path: str) -> None:
+        """Validate and load a file into the session state."""
+        from pathlib import Path as _Path
+        ext = _Path(file_path).suffix.lower()
+        if ext not in SUPPORTED_EXTENSIONS:
+            self._state.reset_file_state()
+            self._refresh_drop_zone(error=True, err_msg=f"Unsupported file type: {ext}")
+            self._page.update()
+            return
+
+        ftype = detect_file_type(file_path)
+        s = self._state
+
+        if ftype in ("epub", "pdf", "markdown"):
+            # For book types: extract text to temp cache
+            self._handle_book_file(file_path, ftype)
+        else:
+            # Plain text / subtitle files
+            s.selected_file = file_path
+            s.selected_file_type = ftype
+            s.displayed_file_path = file_path
+            try:
+                with open(file_path, "r", encoding="utf-8", errors="ignore") as f:
+                    text = f.read()
+                s.char_count = calculate_text_length(clean_text(text))
+            except Exception:
+                s.char_count = 0
+            self._refresh_drop_zone(accent=True)
+            self._update_subtitle_availability()
+            self._page.update()
+
+    def _handle_book_file(self, book_path: str, ftype: str) -> None:
+        """Extract text from epub/pdf/markdown and store as temp txt."""
+        import threading as _t
+        s = self._state
+
+        def _extract():
+            try:
+                from abogen.text_extractor import extract_from_path
+                chapters = extract_from_path(book_path, file_type=ftype)
+                combined = "\n\n".join(ch.text for ch in chapters if ch.text.strip())
+                cache_dir = get_user_cache_path()
+                base = os.path.splitext(os.path.basename(book_path))[0]
+                fd, tmp = tempfile.mkstemp(prefix=f"{base}_", suffix=".txt", dir=cache_dir)
+                os.close(fd)
+                with open(tmp, "w", encoding="utf-8") as f:
+                    f.write(combined)
+
+                s.selected_file = tmp
+                s.selected_file_type = ftype
+                s.selected_book_path = book_path
+                s.displayed_file_path = book_path
+                s.char_count = calculate_text_length(clean_text(combined))
+                s.selected_chapters = [f"ch_{i}" for i in range(len(chapters))]
+
+                self._refresh_drop_zone(accent=True)
+                self._update_subtitle_availability()
+                self._page.update()
+            except Exception as ex:
+                s.reset_file_state()
+                self._refresh_drop_zone(error=True, err_msg=f"Could not parse file: {ex}")
+                self._page.update()
+
+        _t.Thread(target=_extract, daemon=True).start()
+
+    # ------------------------------------------------------------------
+    # Control event handlers
+    # ------------------------------------------------------------------
+
+    def _set_field(self, attr: str, value) -> None:
+        setattr(self._state, attr, value)
+        self._state.persist_config()
+
+    def _on_voice_changed(self, e: ft.ControlEvent) -> None:
+        v = e.control.value or "af_heart"
+        self._state.selected_voice = v
+        self._state.selected_lang = voice_lang_code(v)
+        self._state.persist_config()
+        self._update_subtitle_availability()
+        self._page.update()
+
+    def _on_speed_changed(self, e: ft.ControlEvent) -> None:
+        val = round(float(e.control.value), 2)
+        self._state.speed = val
+        if self._speed_label:
+            self._speed_label.value = f"{val:.2f}"
+        self._state.persist_config()
+        self._page.update()
+
+    def _update_subtitle_availability(self) -> None:
+        """Enable or disable subtitle controls based on selected language."""
+        lang = self._state.selected_lang
+        enabled = lang in SUPPORTED_LANGUAGES_FOR_SUBTITLE_GENERATION
+        if self._subtitle_dd:
+            self._subtitle_dd.disabled = not enabled
+        if self._subtitle_fmt_dd:
+            self._subtitle_fmt_dd.disabled = not enabled
+
+    # ------------------------------------------------------------------
+    # Conversion control
+    # ------------------------------------------------------------------
+
+    def _on_start(self, _: ft.ControlEvent) -> None:
+        """Validate inputs and kick off conversion."""
+        s = self._state
+        if not s.selected_file or not os.path.exists(s.selected_file):
+            self._refresh_drop_zone(error=True, err_msg="Please select an input file first.")
+            self._page.update()
+            return
+
+        # Transition UI to converting state
+        self._set_converting_ui(True)
+
+        self._bridge.start(
+            input_file=s.selected_file,
+            voice=s.get_voice_formula(),
+            lang_code=s.selected_lang,
+            speed=s.speed,
+            output_format=s.selected_format,
+            subtitle_mode=s.subtitle_mode,
+            subtitle_format=s.subtitle_format,
+            use_gpu=s.use_gpu,
+            save_option=s.save_option,
+            output_folder=s.selected_output_folder,
+            replace_single_newlines=s.replace_single_newlines,
+            char_count=s.char_count,
+            save_chapters_separately=s.save_chapters_separately or False,
+            merge_chapters_at_end=True if s.merge_chapters_at_end is None else s.merge_chapters_at_end,
+            separate_chapters_format=s.separate_chapters_format,
+            silence_between_chapters=s.silence_duration,
+            max_subtitle_words=s.max_subtitle_words,
+            chapter_intro_delay=s.chapter_intro_delay,
+            read_title_intro=s.read_title_intro,
+            read_closing_outro=s.read_closing_outro,
+            auto_prefix_chapter_titles=s.auto_prefix_chapter_titles,
+            normalize_chapter_opening_caps=s.normalize_chapter_opening_caps,
+            tts_provider=s.tts_provider,
+            supertonic_total_steps=s.supertonic_total_steps,
+            chunk_level=s.chunk_level,
+            generate_epub3=s.generate_epub3,
+            word_substitutions_enabled=s.word_substitutions_enabled,
+            word_substitutions_list=s.word_substitutions_list,
+            case_sensitive_substitutions=s.case_sensitive_substitutions,
+            replace_all_caps=s.replace_all_caps,
+            replace_numerals=s.replace_numerals,
+            fix_nonstandard_punctuation=s.fix_nonstandard_punctuation,
+        )
+
+    def _on_cancel(self, _: ft.ControlEvent) -> None:
+        self._bridge.cancel()
+
+    def _set_converting_ui(self, converting: bool) -> None:
+        """Toggle UI between idle and converting states."""
+        if self._start_btn:
+            self._start_btn.visible = not converting
+        if self._cancel_btn:
+            self._cancel_btn.visible = converting
+        if self._controls_col:
+            self._controls_col.visible = not converting
+        if self._log_section:
+            self._log_section.visible = converting
+            if self._log_list:
+                self._log_list.controls.clear()
+        if self._progress_col:
+            self._progress_col.visible = converting
+            if self._progress_bar:
+                self._progress_bar.value = 0
+            if self._etr_label:
+                self._etr_label.value = "Estimating…"
+        if self._finish_col:
+            self._finish_col.visible = False
+        self._page.update()
+
+    # ------------------------------------------------------------------
+    # State callbacks (called from background thread via page.run_task)
+    # ------------------------------------------------------------------
+
+    def _on_log(self, message: str, level: str) -> None:
+        if self._log_list is None:
+            return
+        entry = log_entry(message, level, self._page)
+        self._log_list.controls.append(entry)
+        # Cap log lines
+        if len(self._log_list.controls) > 2000:
+            self._log_list.controls = self._log_list.controls[-1800:]
+        try:
+            self._page.update()
+        except Exception:
+            pass
+
+    def _on_progress(self, fraction: float, etr: Optional[float]) -> None:
+        if self._progress_bar:
+            self._progress_bar.value = min(fraction, 0.99)
+        if self._etr_label:
+            self._etr_label.value = format_etr(etr)
+        try:
+            self._page.update()
+        except Exception:
+            pass
+
+    def _on_finished(self, message: str, output_path: Optional[str]) -> None:
+        if self._progress_bar:
+            self._progress_bar.value = 1.0
+        if self._cancel_btn:
+            self._cancel_btn.visible = False
+
+        if message == "Cancelled":
+            # Restore idle state
+            self._set_converting_ui(False)
+            show_snack(self._page, "Conversion cancelled.", error=True)
+            return
+
+        if "failed" in message.lower() or "error" in message.lower():
+            self._log_on_log(message, "error")
+            self._set_converting_ui(False)
+            show_snack(self._page, f"Error: {message}", error=True)
+            return
+
+        # Success
+        if self._log_section:
+            self._log_section.visible = True
+        if self._progress_col:
+            self._progress_col.visible = False
+        if self._controls_col:
+            self._controls_col.visible = False
+        if self._finish_col:
+            self._finish_col.visible = True
+        if self._start_btn:
+            self._start_btn.visible = False
+        show_snack(self._page, "Conversion completed!")
+        try:
+            self._page.update()
+        except Exception:
+            pass
+
+    def _log_on_log(self, message: str, level: str) -> None:
+        self._on_log(message, level)
+
+    # ------------------------------------------------------------------
+    # Finish actions
+    # ------------------------------------------------------------------
+
+    def _on_open_file(self, _: ft.ControlEvent) -> None:
+        path = self._state.last_output_path
+        if path and os.path.exists(path):
+            import subprocess, platform
+            try:
+                if platform.system() == "Darwin":
+                    subprocess.Popen(["open", path])
+                elif platform.system() == "Windows":
+                    os.startfile(path)
+                else:
+                    subprocess.Popen(["xdg-open", path])
+            except Exception as ex:
+                show_snack(self._page, f"Cannot open file: {ex}", error=True)
+        else:
+            show_snack(self._page, "Output file not found.", error=True)
+
+    def _on_go_folder(self, _: ft.ControlEvent) -> None:
+        path = self._state.last_output_path
+        folder = os.path.dirname(path) if path and os.path.isfile(path) else path
+        if folder and os.path.isdir(folder):
+            import subprocess, platform
+            try:
+                if platform.system() == "Darwin":
+                    subprocess.Popen(["open", folder])
+                elif platform.system() == "Windows":
+                    subprocess.Popen(["explorer", folder])
+                else:
+                    subprocess.Popen(["xdg-open", folder])
+            except Exception as ex:
+                show_snack(self._page, f"Cannot open folder: {ex}", error=True)
+        else:
+            show_snack(self._page, "Output folder not found.", error=True)
+
+    def _on_reset(self, _: ft.ControlEvent) -> None:
+        self._state.reset_file_state()
+        self._state.reset_conversion_state()
+        self._refresh_drop_zone()
+        self._set_converting_ui(False)
+        if self._finish_col:
+            self._finish_col.visible = False
+        if self._controls_col:
+            self._controls_col.visible = True
+        if self._start_btn:
+            self._start_btn.visible = True
+        self._page.update()

abogen/frontend/views/queue_view.py

@@ -0,0 +1,154 @@
+"""
+Queue management view.
+
+Displays the current conversion queue, allowing the user to reorder,
+remove, and inspect queued items before starting batch processing.
+"""
+from __future__ import annotations
+from typing import Optional
+
+import flet as ft
+
+from ..state import AppState, ConversionJob
+from ..utils.theme import get_palette, RADIUS_SM, SPACE_SM, SPACE_MD, SPACE_LG
+from ..utils.helpers import safe_basename, output_format_label, format_number
+from ..components import (
+    build_card, build_section_header, build_primary_button,
+    build_secondary_button, show_snack, build_divider,
+    resolve_icon,
+)
+
+
+class QueueView:
+    """Queue manager view."""
+
+    def __init__(self, page: ft.Page, state: AppState) -> None:
+        self._page = page
+        self._state = state
+        self._list_col: Optional[ft.Column] = None
+
+    def build(self) -> ft.Column:
+        p = self._page
+        s = self._state
+        pal = get_palette(p)
+        dark = p.theme_mode == ft.ThemeMode.DARK
+
+        self._list_col = ft.Column(spacing=SPACE_SM)
+        self._refresh_list()
+
+        header = build_section_header("Conversion Queue",
+                                      icon="list_alt", page=p)
+
+        action_row = ft.Row([
+            build_primary_button(
+                "Start Queue",
+                icon="play_arrow",
+                on_click=self._on_start_queue,
+                page=p,
+                disabled=not s.queued_items,
+            ),
+            build_secondary_button(
+                "Clear All",
+                icon="delete_sweep",
+                on_click=self._on_clear_queue,
+                page=p,
+            ),
+        ], spacing=SPACE_SM, wrap=True)
+
+        queue_card = build_card(ft.Column([
+            header,
+            ft.Divider(height=1, color=pal.divider),
+            self._list_col,
+            ft.Container(height=SPACE_SM),
+            action_row,
+        ], spacing=SPACE_MD), page=p)
+
+        return ft.Column([queue_card], scroll=ft.ScrollMode.AUTO, expand=True)
+
+    # ------------------------------------------------------------------
+
+    def _refresh_list(self) -> None:
+        if self._list_col is None:
+            return
+        self._list_col.controls.clear()
+        s = self._state
+        pal = get_palette(self._page)
+        dark = self._page.theme_mode == ft.ThemeMode.DARK
+
+        if not s.queued_items:
+            self._list_col.controls.append(
+                ft.Text("No items in the queue.", size=13,
+                        color=pal.text_secondary,
+                        text_align=ft.TextAlign.CENTER)
+            )
+            return
+
+        for idx, job in enumerate(s.queued_items):
+            tile = self._build_job_tile(idx, job, dark, pal)
+            self._list_col.controls.append(tile)
+
+        try:
+            self._page.update()
+        except Exception:
+            pass
+
+    def _build_job_tile(self, idx: int, job: ConversionJob, dark: bool, pal) -> ft.Container:
+        """Build a single queue-item tile."""
+        bg = pal.bg_elevated
+        border_clr = pal.border
+        accent = "#5b8af5" if dark else "#3a5fc4"
+        text_primary = pal.text_primary
+        text_secondary = pal.text_secondary
+
+        def _remove(_):
+            self._state.queued_items.pop(idx)
+            self._refresh_list()
+
+        name = safe_basename(job.display_name or job.file_path)
+        details = (
+            f"Voice: {job.voice}  ·  Format: {output_format_label(job.output_format)}"
+            f"  ·  Speed: {job.speed:.2f}x  ·  Chars: {format_number(job.char_count)}"
+        )
+
+        return ft.Container(
+            content=ft.Row([
+                ft.Container(
+                    content=ft.Text(str(idx + 1), size=12, weight=ft.FontWeight.W_700,
+                                    color=accent),
+                    width=32,
+                ),
+                ft.Column([
+                    ft.Text(name, size=13, weight=ft.FontWeight.W_600, color=text_primary,
+                            no_wrap=True, overflow=ft.TextOverflow.ELLIPSIS),
+                    ft.Text(details, size=11, color=text_secondary),
+                ], expand=True, tight=True, spacing=2),
+                ft.IconButton(
+                    icon=resolve_icon("delete_outline"),
+                    icon_color=pal.error if hasattr(pal, "error") else "#e84e3c",
+                    icon_size=18,
+                    tooltip="Remove",
+                    on_click=_remove,
+                ),
+            ], vertical_alignment=ft.CrossAxisAlignment.CENTER, spacing=SPACE_SM),
+            bgcolor=bg,
+            border=ft.Border.all(1, border_clr),
+            border_radius=RADIUS_SM,
+            padding=ft.Padding.symmetric(horizontal=SPACE_MD, vertical=SPACE_SM),
+        )
+
+    # ------------------------------------------------------------------
+
+    def _on_start_queue(self, _: ft.ControlEvent) -> None:
+        if not self._state.queued_items:
+            show_snack(self._page, "Queue is empty.", error=True)
+            return
+        # Navigate to dashboard and trigger queue start
+        # This is wired in main.py via the nav controller
+        self._page.pubsub.send_all("start_queue")
+
+    def _on_clear_queue(self, _: ft.ControlEvent) -> None:
+        if not self._state.queued_items:
+            return
+        self._state.queued_items.clear()
+        self._refresh_list()
+        show_snack(self._page, "Queue cleared.")

abogen/frontend/views/settings.py

@@ -0,0 +1,305 @@
+"""
+Settings view – a categorised, scrollable settings page.
+
+Groups settings into collapsible cards:
+  - Output (format, save location, chapters)
+  - Text processing (newlines, caps, substitutions, numerals)
+  - Subtitle options
+  - TTS pipeline (provider, GPU, chunking)
+  - Integrations (Audiobookshelf, Calibre OPDS)
+"""
+from __future__ import annotations
+from typing import Optional
+
+import flet as ft
+
+from ..state import AppState
+from ..utils.theme import get_palette, RADIUS_MD, RADIUS_SM, SPACE_SM, SPACE_MD, SPACE_LG
+from ..utils.helpers import output_format_label, subtitle_format_label, SUPPORTED_EXTENSIONS
+from ..components import (
+    build_card, build_section_header, labelled_row, show_snack, build_divider,
+    build_primary_button,
+)
+from abogen.constants import SUBTITLE_FORMATS
+
+
+# ---------------------------------------------------------------------------
+# Helpers
+# ---------------------------------------------------------------------------
+
+def _dd(options, value, on_change, **kw):
+    """Compact dropdown factory."""
+    return ft.Dropdown(
+        options=[ft.dropdown.Option(key=k, text=v) for k, v in options],
+        value=value, on_select=on_change, dense=True,
+        border_radius=RADIUS_SM, expand=True, **kw
+    )
+
+
+def _sw(value, on_change, label=""):
+    return ft.Switch(value=value, on_change=on_change, label=label)
+
+
+class SettingsView:
+    """The full settings panel."""
+
+    def __init__(self, page: ft.Page, state: AppState) -> None:
+        self._page = page
+        self._state = state
+
+    def build(self) -> ft.Column:
+        p = self._page
+        s = self._state
+        pal = get_palette(p)
+
+        # ── Output card ──────────────────────────────────────────────
+        format_dd = _dd(
+            [(k, output_format_label(k)) for k in ("wav", "flac", "mp3", "opus", "m4b")],
+            s.selected_format,
+            lambda e: self._save("selected_format", e.control.value),
+        )
+        save_dd = _dd(
+            [
+                ("Save next to input file", "Save next to input file"),
+                ("Save to Desktop", "Save to Desktop"),
+                ("Choose output folder", "Choose output folder"),
+            ],
+            s.save_option,
+            lambda e: self._save("save_option", e.control.value),
+        )
+        chapters_sw = _sw(s.save_chapters_separately or False,
+                          lambda e: self._save("save_chapters_separately", e.control.value))
+        merge_sw = _sw(True if s.merge_chapters_at_end is None else s.merge_chapters_at_end,
+                       lambda e: self._save("merge_chapters_at_end", e.control.value))
+        sep_fmt_dd = _dd(
+            [(k, output_format_label(k)) for k in ("wav", "flac", "mp3", "opus")],
+            s.separate_chapters_format,
+            lambda e: self._save("separate_chapters_format", e.control.value),
+        )
+        epub3_sw = _sw(s.generate_epub3, lambda e: self._save("generate_epub3", e.control.value))
+
+        output_card = build_card(ft.Column([
+            build_section_header("Output", icon="audio_file", page=p),
+            labelled_row("Audio Format", format_dd, page=p),
+            labelled_row("Save Location", save_dd, page=p),
+            build_divider(p),
+            labelled_row("Save Chapters Separately", chapters_sw, page=p),
+            labelled_row("Merge at End", merge_sw, page=p),
+            labelled_row("Chapter Format", sep_fmt_dd, page=p),
+            labelled_row("Generate EPUB3", epub3_sw, page=p),
+        ], spacing=SPACE_MD), page=p)
+
+        # ── Text processing card ─────────────────────────────────────
+        newlines_sw = _sw(s.replace_single_newlines,
+                          lambda e: self._save("replace_single_newlines", e.control.value))
+        caps_sw = _sw(s.replace_all_caps, lambda e: self._save("replace_all_caps", e.control.value))
+        norm_sw = _sw(s.normalize_chapter_opening_caps,
+                      lambda e: self._save("normalize_chapter_opening_caps", e.control.value))
+        numerals_sw = _sw(s.replace_numerals, lambda e: self._save("replace_numerals", e.control.value))
+        punct_sw = _sw(s.fix_nonstandard_punctuation,
+                       lambda e: self._save("fix_nonstandard_punctuation", e.control.value))
+        wordsub_sw = _sw(s.word_substitutions_enabled,
+                         lambda e: self._save("word_substitutions_enabled", e.control.value))
+        wordsub_tf = ft.TextField(
+            value=s.word_substitutions_list,
+            multiline=True, min_lines=3, max_lines=6,
+            hint_text="word|replacement  (one per line)",
+            on_change=lambda e: self._save("word_substitutions_list", e.control.value),
+            expand=True, border_radius=RADIUS_SM, text_size=12,
+        )
+        case_sw = _sw(s.case_sensitive_substitutions,
+                      lambda e: self._save("case_sensitive_substitutions", e.control.value))
+        spacy_sw = _sw(s.use_spacy_segmentation,
+                       lambda e: self._save("use_spacy_segmentation", e.control.value))
+        chunk_dd = _dd(
+            [("paragraph", "Paragraph"), ("sentence", "Sentence")],
+            s.chunk_level,
+            lambda e: self._save("chunk_level", e.control.value),
+        )
+        title_intro_sw = _sw(s.read_title_intro, lambda e: self._save("read_title_intro", e.control.value))
+        outro_sw = _sw(s.read_closing_outro, lambda e: self._save("read_closing_outro", e.control.value))
+        prefix_sw = _sw(s.auto_prefix_chapter_titles,
+                        lambda e: self._save("auto_prefix_chapter_titles", e.control.value))
+
+        text_card = build_card(ft.Column([
+            build_section_header("Text Processing", icon="text_fields", page=p),
+            labelled_row("Replace Single Newlines", newlines_sw,
+                         tooltip="Replace single newlines with spaces before processing.", page=p),
+            labelled_row("Replace ALL CAPS Words", caps_sw, page=p),
+            labelled_row("Normalize Opening CAPS", norm_sw, page=p),
+            labelled_row("Replace Numerals (spoken)", numerals_sw, page=p),
+            labelled_row("Fix Non-standard Punctuation", punct_sw, page=p),
+            build_divider(p),
+            labelled_row("Word Substitutions", wordsub_sw, page=p),
+            labelled_row("Case Sensitive", case_sw, page=p),
+            ft.Text("Substitution rules  (word|replacement, one per line):",
+                    size=12, color=pal.text_secondary),
+            wordsub_tf,
+            build_divider(p),
+            build_section_header("Chapter Options", icon="library_books", page=p),
+            labelled_row("Announce Book Title (intro)", title_intro_sw, page=p),
+            labelled_row("Announce Book Title (outro)", outro_sw, page=p),
+            labelled_row("Auto-prefix Chapter Titles", prefix_sw, page=p),
+            labelled_row("Chunk Level", chunk_dd, page=p),
+            labelled_row("Use spaCy Segmentation", spacy_sw, page=p),
+        ], spacing=SPACE_MD), page=p)
+
+        # ── Subtitle card ─────────────────────────────────────────────
+        sub_modes = ["Disabled", "Line", "Sentence", "Sentence + Comma",
+                     "Sentence + Highlighting"] + [f"{i} word{'s' if i > 1 else ''}" for i in range(1, 11)]
+        sub_mode_dd = _dd(
+            [(m, m) for m in sub_modes],
+            s.subtitle_mode,
+            lambda e: self._save("subtitle_mode", e.control.value),
+        )
+        sub_fmt_dd = _dd(
+            [(k, lbl) for k, lbl in SUBTITLE_FORMATS],
+            s.subtitle_format,
+            lambda e: self._save("subtitle_format", e.control.value),
+        )
+
+        def _mk_mw_slider():
+            lbl = ft.Text(str(s.max_subtitle_words), size=12, width=36)
+            sl = ft.Slider(
+                min=1, max=200, value=s.max_subtitle_words, divisions=199, label="{value}",
+                expand=True,
+                on_change=lambda e: (self._save("max_subtitle_words", int(e.control.value)),
+                                     setattr(lbl, "value", str(int(e.control.value))),
+                                     self._page.update()),
+            )
+            return ft.Row([sl, lbl], expand=True, spacing=SPACE_SM)
+
+        sub_speed_dd = _dd(
+            [("tts", "TTS duration"), ("silence", "Silence detection")],
+            s.subtitle_speed_method,
+            lambda e: self._save("subtitle_speed_method", e.control.value),
+        )
+        silent_gaps_sw = _sw(s.use_silent_gaps,
+                             lambda e: self._save("use_silent_gaps", e.control.value))
+
+        subtitle_card = build_card(ft.Column([
+            build_section_header("Subtitles", icon="subtitles", page=p),
+            labelled_row("Mode", sub_mode_dd, page=p),
+            labelled_row("Format", sub_fmt_dd, page=p),
+            labelled_row("Max Words / Block", _mk_mw_slider(), page=p),
+            labelled_row("Speed Method", sub_speed_dd, page=p),
+            labelled_row("Silent Gaps", silent_gaps_sw, page=p),
+        ], spacing=SPACE_MD), page=p)
+
+        # ── Pipeline card ─────────────────────────────────────────────
+        provider_dd = _dd(
+            [("kokoro", "Kokoro (default)"), ("supertonic", "Supertonic")],
+            s.tts_provider,
+            lambda e: self._save("tts_provider", e.control.value),
+        )
+        gpu_sw = _sw(s.use_gpu, lambda e: self._save("use_gpu", e.control.value),
+                     label="GPU acceleration (if available)")
+
+        def _mk_steps_slider():
+            lbl = ft.Text(str(s.supertonic_total_steps), size=12, width=28)
+            sl = ft.Slider(
+                min=2, max=15, value=s.supertonic_total_steps, divisions=13,
+                label="{value}", expand=True,
+                on_change=lambda e: (self._save("supertonic_total_steps", int(e.control.value)),
+                                     setattr(lbl, "value", str(int(e.control.value))),
+                                     self._page.update()),
+            )
+            return ft.Row([sl, lbl], expand=True, spacing=SPACE_SM)
+
+        thresh_tf = ft.TextField(
+            value=str(s.speaker_analysis_threshold), width=80,
+            keyboard_type=ft.KeyboardType.NUMBER, border_radius=RADIUS_SM,
+            on_change=lambda e: self._save_int("speaker_analysis_threshold", e.control.value, 1, 25),
+        )
+        silence_tf = ft.TextField(
+            value=str(s.silence_duration), width=80,
+            keyboard_type=ft.KeyboardType.NUMBER, border_radius=RADIUS_SM,
+            on_change=lambda e: self._save_float("silence_duration", e.control.value, 0.0),
+        )
+        intro_tf = ft.TextField(
+            value=str(s.chapter_intro_delay), width=80,
+            keyboard_type=ft.KeyboardType.NUMBER, border_radius=RADIUS_SM,
+            on_change=lambda e: self._save_float("chapter_intro_delay", e.control.value, 0.0),
+        )
+
+        pipeline_card = build_card(ft.Column([
+            build_section_header("TTS Pipeline", icon="settings", page=p),
+            labelled_row("Provider", provider_dd, page=p),
+            labelled_row("GPU Acceleration", gpu_sw, page=p),
+            labelled_row("Supertonic Steps", _mk_steps_slider(), page=p),
+            build_divider(p),
+            labelled_row("Speaker Analysis Threshold", thresh_tf, page=p),
+            labelled_row("Silence Between Chapters (s)", silence_tf, page=p),
+            labelled_row("Chapter Intro Delay (s)", intro_tf, page=p),
+        ], spacing=SPACE_MD), page=p)
+
+        # ── Integration card (Audiobookshelf) ─────────────────────────
+        abs_enabled_sw = _sw(s.audiobookshelf_enabled,
+                             lambda e: self._save("audiobookshelf_enabled", e.control.value))
+        abs_url_tf = ft.TextField(value=s.audiobookshelf_base_url, hint_text="http://abs-server:13378",
+                                  expand=True, border_radius=RADIUS_SM, text_size=12,
+                                  on_change=lambda e: self._save("audiobookshelf_base_url", e.control.value))
+        abs_token_tf = ft.TextField(value=s.audiobookshelf_api_token, password=True,
+                                    can_reveal_password=True, expand=True,
+                                    border_radius=RADIUS_SM, text_size=12,
+                                    on_change=lambda e: self._save("audiobookshelf_api_token", e.control.value))
+        abs_lib_tf = ft.TextField(value=s.audiobookshelf_library_id, hint_text="Library ID",
+                                  expand=True, border_radius=RADIUS_SM, text_size=12,
+                                  on_change=lambda e: self._save("audiobookshelf_library_id", e.control.value))
+        abs_auto_sw = _sw(s.audiobookshelf_auto_send,
+                          lambda e: self._save("audiobookshelf_auto_send", e.control.value))
+
+        integ_card = build_card(ft.Column([
+            build_section_header("Audiobookshelf Integration",
+                                 icon="cloud_upload", page=p),
+            labelled_row("Enabled", abs_enabled_sw, page=p),
+            labelled_row("Server URL", abs_url_tf, page=p),
+            labelled_row("API Token", abs_token_tf, page=p),
+            labelled_row("Library ID", abs_lib_tf, page=p),
+            labelled_row("Auto-upload on finish", abs_auto_sw, page=p),
+        ], spacing=SPACE_MD), page=p)
+
+        save_btn = build_primary_button(
+            "Save Settings", icon="save",
+            on_click=self._on_save, page=p,
+        )
+
+        return ft.Column([
+            output_card,
+            ft.Container(height=SPACE_MD),
+            text_card,
+            ft.Container(height=SPACE_MD),
+            subtitle_card,
+            ft.Container(height=SPACE_MD),
+            pipeline_card,
+            ft.Container(height=SPACE_MD),
+            integ_card,
+            ft.Container(height=SPACE_LG),
+            save_btn,
+            ft.Container(height=SPACE_LG),
+        ], spacing=0, scroll=ft.ScrollMode.AUTO, expand=True)
+
+    # ------------------------------------------------------------------
+    # Helpers
+    # ------------------------------------------------------------------
+
+    def _save(self, attr: str, value) -> None:
+        setattr(self._state, attr, value)
+
+    def _save_int(self, attr: str, raw: str, lo: int, hi: int) -> None:
+        try:
+            v = max(lo, min(hi, int(raw)))
+            setattr(self._state, attr, v)
+        except ValueError:
+            pass
+
+    def _save_float(self, attr: str, raw: str, lo: float) -> None:
+        try:
+            v = max(lo, float(raw))
+            setattr(self._state, attr, v)
+        except ValueError:
+            pass
+
+    def _on_save(self, _: ft.ControlEvent) -> None:
+        self._state.persist_config()
+        show_snack(self._page, "Settings saved.")

abogen/pyqt/conversion.py

@@ -42,6 +42,10 @@ from abogen.subtitle_utils import (
     get_sample_voice_text,
     sanitize_name_for_os,
     _CHAPTER_MARKER_SEARCH_PATTERN,
+    _VOICE_MARKER_PATTERN,
+    _VOICE_MARKER_SEARCH_PATTERN,
+    split_text_by_voice_markers,
+    validate_voice_name,
 )
 
 class CountdownDialog(QDialog):
@@ -296,6 +300,31 @@ class ConversionThread(QThread):
         self.use_spacy_segmentation = True  # Default, will be overridden from GUI
         # Set split pattern based on language and subtitle mode
         self.split_pattern = self._get_split_pattern(lang_code, subtitle_mode)
+        self.voice_cache = {}  # Cache for loaded voices
+
+    def load_voice_cached(self, voice_name, tts):
+        """Load voice with caching to avoid reloading same voice.
+
+        Args:
+            voice_name: Voice name or formula string
+            tts: TTS pipeline instance
+
+        Returns:
+            Loaded voice tensor or voice name string
+        """
+        # Check cache first
+        if voice_name in self.voice_cache:
+            return self.voice_cache[voice_name]
+
+        # Load voice
+        if "*" in voice_name:
+            loaded_voice = get_new_voice(tts, voice_name, self.use_gpu)
+        else:
+            loaded_voice = voice_name
+
+        # Cache it
+        self.voice_cache[voice_name] = loaded_voice
+        return loaded_voice
 
     def _stream_audio_in_chunks(
         self, segments, process_func, progress_prefix="Processing"
@@ -524,6 +553,26 @@ class ConversionThread(QThread):
             # Clean up text using utility function
             text = clean_text(text)
 
+            # Apply word substitutions if enabled
+            if getattr(self, "word_substitutions_enabled", False):
+                from abogen.word_substitution import apply_word_substitutions
+
+                self.log_updated.emit("Applying word substitutions...")
+
+                substitutions_list = getattr(self, "word_substitutions_list", "")
+                case_sensitive = getattr(self, "case_sensitive_substitutions", False)
+                replace_caps = getattr(self, "replace_all_caps", False)
+                replace_nums = getattr(self, "replace_numerals", False)
+                fix_punct = getattr(self, "fix_nonstandard_punctuation", False)
+
+                text = apply_word_substitutions(
+                    text,
+                    substitutions_list,
+                    case_sensitive,
+                    replace_caps,
+                    replace_nums,
+                    fix_punct,
+                )
 
             # --- Chapter splitting logic ---
             # Use pre-compiled pattern for better performance
@@ -550,6 +599,42 @@ class ConversionThread(QThread):
                 chapters = [("text", text)]
             total_chapters = len(chapters)
 
+            # --- Voice marker splitting logic ---
+            # Split each chapter by voice markers, preserving voice state across chapters
+            chapters_with_voices = []
+            current_voice = self.voice  # Start with default voice
+            total_valid_markers = 0
+            total_invalid_markers = 0
+
+            for chapter_name, chapter_text in chapters:
+                # Use current_voice as the starting voice for this chapter
+                voice_segments, last_voice, valid_count, invalid_count = split_text_by_voice_markers(chapter_text, current_voice)
+                chapters_with_voices.append((chapter_name, voice_segments))
+
+                # Update current_voice so next chapter continues with this voice
+                current_voice = last_voice
+
+                # Track total valid/invalid markers
+                total_valid_markers += valid_count
+                total_invalid_markers += invalid_count
+
+            # Log voice marker information with accurate counts
+            total_markers = total_valid_markers + total_invalid_markers
+            if total_markers > 0:
+                if total_invalid_markers == 0:
+                    # All markers were valid
+                    self.log_updated.emit(
+                        (f"\nDetected {total_markers} voice marker(s) - all valid", "grey")
+                    )
+                else:
+                    # Some markers were invalid
+                    self.log_updated.emit(
+                        (f"\nDetected {total_markers} voice marker(s) - {total_valid_markers} valid, {total_invalid_markers} invalid (using previous voice)", "orange")
+                    )
+
+            # Replace chapters with the new structure
+            chapters = chapters_with_voices
+
             # For text files with chapters, prompt user for options if not already set
             is_txt_file = not self.is_direct_text and (
                 self.file_name.lower().endswith(".txt")
@@ -842,7 +927,7 @@ class ConversionThread(QThread):
                 ]
                 srt_index = 1  # SRT numbering fix for chapter-only mode
             # Instead of processing the whole text, process by chapter
-            for chapter_idx, (chapter_name, chapter_text) in enumerate(chapters, 1):
+            for chapter_idx, (chapter_name, voice_segments) in enumerate(chapters, 1):
                 chapter_out_path = None
                 chapter_out_file = None
                 chapter_ffmpeg_proc = None
@@ -862,11 +947,6 @@ class ConversionThread(QThread):
                 if merge_chapters_at_end:
                     chapter_time["start"] = current_time
 
-                # Check if the voice is a formula and load it if necessary
-                if "*" in self.voice:
-                    loaded_voice = get_new_voice(tts, self.voice, self.use_gpu)
-                else:
-                    loaded_voice = self.voice
                 # Prepare per-chapter output file if needed
                 if save_chapters_separately and total_chapters > 1:
                     # First pass: keep alphanumeric, spaces, hyphens, and underscores
@@ -986,286 +1066,302 @@ class ConversionThread(QThread):
                     chapter_subtitle_path = None
                     chapter_subtitle_file = None
 
-                # Determine if spaCy segmentation should be used for PRE-TTS segmentation
-                # Only non-English languages use spaCy for pre-segmentation
-                # English uses spaCy only for subtitle generation (post-TTS)
-                # spaCy is disabled when subtitle mode is "Disabled" or "Line"
-                # spaCy is also disabled when input is a subtitle file
-                is_subtitle_input = (
-                    not self.is_direct_text
-                    and self.file_name
-                    and os.path.splitext(self.file_name)[1].lower()
-                    in [".srt", ".ass", ".vtt"]
-                )
-                use_spacy = (
-                    getattr(self, "use_spacy_segmentation", False)
-                    and self.subtitle_mode not in ["Disabled", "Line"]
-                    and not is_subtitle_input
-                )
-                spacy_sentences = None
-                active_split_pattern = self.split_pattern
-                spacing_pattern = r"\s*" if self.lang_code in ["z", "j"] else r"\s+"
 
-                # Pre-load spaCy model for English if it will be needed for subtitle generation
+                # Process each voice segment within the chapter
-                if (
+                for segment_idx, (voice_name, segment_text) in enumerate(voice_segments):
-                    use_spacy
+                    # Load voice for this segment (with caching)
-                    and self.lang_code in ["a", "b"]
+                    try:
-                    and self.subtitle_mode in ["Sentence", "Sentence + Comma"]
+                        loaded_voice = self.load_voice_cached(voice_name, tts)
-                ):
+                        if segment_idx > 0:
-                    from abogen.spacy_utils import get_spacy_model
+                            voice_display = voice_name if len(voice_name) < 50 else voice_name[:47] + "..."
-
+                            self.log_updated.emit((f"  → Voice: {voice_display}", "grey"))
-                    nlp = get_spacy_model(
+                    except Exception:
-                        self.lang_code,
-                        log_callback=lambda msg: self.log_updated.emit(msg),
-                    )
-                    if nlp:
                         self.log_updated.emit(
-                            (
+                            (f"⚠ Voice loading error for '{voice_name}', continuing with previous", "orange")
-                                "\nUsing spaCy for sentence segmentation (only for subtitles)...",
-                                "grey",
-                            )
                         )
+                        if segment_idx == 0:
+                            loaded_voice = self.load_voice_cached(self.voice, tts)
 
-                if use_spacy and self.lang_code not in ["a", "b"]:
+                    # Determine if spaCy segmentation should be used for PRE-TTS segmentation
-                    # Non-English: use spaCy for pre-TTS segmentation
+                    # Only non-English languages use spaCy for pre-segmentation
-                    self.log_updated.emit(
+                    # English uses spaCy only for subtitle generation (post-TTS)
-                        ("\nUsing spaCy for sentence segmentation (pre-TTS)...", "grey")
+                    # spaCy is disabled when subtitle mode is "Disabled" or "Line"
+                    # spaCy is also disabled when input is a subtitle file
+                    is_subtitle_input = (
+                        not self.is_direct_text
+                        and self.file_name
+                        and os.path.splitext(self.file_name)[1].lower()
+                        in [".srt", ".ass", ".vtt"]
                     )
-                    from abogen.spacy_utils import segment_sentences
+                    use_spacy = (
-
+                        getattr(self, "use_spacy_segmentation", False)
-                    spacy_sentences = segment_sentences(
+                        and self.subtitle_mode not in ["Disabled", "Line"]
-                        chapter_text,
+                        and not is_subtitle_input
-                        self.lang_code,
-                        log_callback=lambda msg: self.log_updated.emit(msg),
                     )
-                    if spacy_sentences:
+                    spacy_sentences = None
-                        self.log_updated.emit(
+                    active_split_pattern = self.split_pattern
-                            (
+                    spacing_pattern = r"\s*" if self.lang_code in ["z", "j"] else r"\s+"
-                                f"\nspaCy: Text segmented into {len(spacy_sentences)} sentences...",
-                                "grey",
-                            )
-                        )
-                        # For Sentence + Comma mode, still split on commas within spaCy sentences
-                        if self.subtitle_mode == "Sentence + Comma":
-                            active_split_pattern = r"(?<=[{}]){}|\n+".format(
-                                self.PUNCTUATION_COMMAS, spacing_pattern
-                            )
-                        else:
-                            active_split_pattern = (
-                                "\n"  # Use newline splitting for Sentence mode
-                            )
-                    else:
-                        self.log_updated.emit(
-                            ("\nspaCy: Fallback to default segmentation...", "grey")
-                        )
 
-                # Process text - either as spaCy sentences or as single text
+                    # Pre-load spaCy model for English if it will be needed for subtitle generation
-                text_segments = spacy_sentences if spacy_sentences else [chapter_text]
+                    if (
-
+                        use_spacy
-                # Print active split pattern used by the TTS engine once for this batch
+                        and self.lang_code in ["a", "b"]
-                try:
+                        and self.subtitle_mode in ["Sentence", "Sentence + Comma"]
-                    print(f"Using split pattern: {active_split_pattern!r}")
-                except Exception:
-                    # Print must never break processing
-                    print("Using split pattern: (unprintable)")
-
-                for text_segment in text_segments:
-                    for result in tts(
-                        text_segment,
-                        voice=loaded_voice,
-                        speed=self.speed,
-                        split_pattern=active_split_pattern,
                     ):
-                        # Print the result for debugging
+                        from abogen.spacy_utils import get_spacy_model
-                        # print(f"Result: {result}")
+
-                        if self.cancel_requested:
+                        nlp = get_spacy_model(
-                            if chapter_out_file:
+                            self.lang_code,
-                                chapter_out_file.close()
+                            log_callback=lambda msg: self.log_updated.emit(msg),
-                            if merged_out_file:
-                                merged_out_file.close()
-                            self.conversion_finished.emit("Cancelled", None)
-                            return
-                        current_segment += 1
-                        grapheme_len = len(result.graphemes)
-                        self.processed_char_count += grapheme_len
-                        # Log progress with both character counts and the graphemes content
-                        self.log_updated.emit(
-                            f"\n{self.processed_char_count:,}/{self.total_char_count:,}: {result.graphemes}"
                         )
+                        if nlp:
+                            self.log_updated.emit(
+                                (
+                                    "\nUsing spaCy for sentence segmentation (only for subtitles)...",
+                                    "grey",
+                                )
+                            )
 
-                        chunk_dur = len(result.audio) / rate
+                    if use_spacy and self.lang_code not in ["a", "b"]:
-                        chunk_start = current_time
+                        # Non-English: use spaCy for pre-TTS segmentation
-                        # Write audio directly to merged file ONLY if merging
+                        self.log_updated.emit(
-                        if merge_chapters_at_end and merged_out_file:
+                            ("\nUsing spaCy for sentence segmentation (pre-TTS)...", "grey")
-                            merged_out_file.write(result.audio)
+                        )
-                        elif merge_chapters_at_end and ffmpeg_proc:
+                        from abogen.spacy_utils import segment_sentences
-                            if hasattr(result.audio, "numpy"):
+
-                                audio_bytes = (
+                        spacy_sentences = segment_sentences(
-                                    result.audio.numpy().astype("float32").tobytes()
+                            segment_text,
+                            self.lang_code,
+                            log_callback=lambda msg: self.log_updated.emit(msg),
+                        )
+                        if spacy_sentences:
+                            self.log_updated.emit(
+                                (
+                                    f"\nspaCy: Text segmented into {len(spacy_sentences)} sentences...",
+                                    "grey",
+                                )
+                            )
+                            # For Sentence + Comma mode, still split on commas within spaCy sentences
+                            if self.subtitle_mode == "Sentence + Comma":
+                                active_split_pattern = r"(?<=[{}]){}|\n+".format(
+                                    self.PUNCTUATION_COMMAS, spacing_pattern
                                 )
                             else:
-                                audio_bytes = result.audio.astype("float32").tobytes()
+                                active_split_pattern = (
-                            ffmpeg_proc.stdin.write(audio_bytes)
+                                    "\n"  # Use newline splitting for Sentence mode
-                        if chapter_out_file:
-                            chapter_out_file.write(result.audio)
-                        elif chapter_ffmpeg_proc:
-                            if hasattr(result.audio, "numpy"):
-                                audio_bytes = (
-                                    result.audio.numpy().astype("float32").tobytes()
                                 )
-                            else:
+                        else:
-                                audio_bytes = result.audio.astype("float32").tobytes()
+                            self.log_updated.emit(
-                            chapter_ffmpeg_proc.stdin.write(audio_bytes)
+                                ("\nspaCy: Fallback to default segmentation...", "grey")
-                        # Subtitle logic
+                            )
-                        if self.subtitle_mode != "Disabled":
-                            tokens_list = getattr(result, "tokens", [])
 
-                            # Fallback for languages without token support (non-English)
+                    # Process text - either as spaCy sentences or as single text
-                            # Create a single token representing the entire segment duration
+                    text_segments = spacy_sentences if spacy_sentences else [segment_text]
-                            if not tokens_list and result.graphemes:
 
-                                class FakeToken:
+                    # Print active split pattern used by the TTS engine once for this batch
-                                    def __init__(self, text, start, end):
+                    try:
-                                        self.text = text
+                        print(f"Using split pattern: {active_split_pattern!r}")
-                                        self.start_ts = start
+                    except Exception:
-                                        self.end_ts = end
+                        # Print must never break processing
-                                        self.whitespace = ""
+                        print("Using split pattern: (unprintable)")
 
-                                tokens_list = [
+                    for text_segment in text_segments:
-                                    FakeToken(result.graphemes, 0, chunk_dur)
+                        for result in tts(
-                                ]
+                            text_segment,
+                            voice=loaded_voice,
+                            speed=self.speed,
+                            split_pattern=active_split_pattern,
+                        ):
+                            # Print the result for debugging
+                            # print(f"Result: {result}")
+                            if self.cancel_requested:
+                                if chapter_out_file:
+                                    chapter_out_file.close()
+                                if merged_out_file:
+                                    merged_out_file.close()
+                                self.conversion_finished.emit("Cancelled", None)
+                                return
+                            current_segment += 1
+                            grapheme_len = len(result.graphemes)
+                            self.processed_char_count += grapheme_len
+                            # Log progress with both character counts and the graphemes content
+                            self.log_updated.emit(
+                                f"\n{self.processed_char_count:,}/{self.total_char_count:,}: {result.graphemes}"
+                            )
 
-                            tokens_with_timestamps = []
+                            chunk_dur = len(result.audio) / rate
-                            chapter_tokens_with_timestamps = []
+                            chunk_start = current_time
+                            # Write audio directly to merged file ONLY if merging
+                            if merge_chapters_at_end and merged_out_file:
+                                merged_out_file.write(result.audio)
+                            elif merge_chapters_at_end and ffmpeg_proc:
+                                if hasattr(result.audio, "numpy"):
+                                    audio_bytes = (
+                                        result.audio.numpy().astype("float32").tobytes()
+                                    )
+                                else:
+                                    audio_bytes = result.audio.astype("float32").tobytes()
+                                ffmpeg_proc.stdin.write(audio_bytes)
+                            if chapter_out_file:
+                                chapter_out_file.write(result.audio)
+                            elif chapter_ffmpeg_proc:
+                                if hasattr(result.audio, "numpy"):
+                                    audio_bytes = (
+                                        result.audio.numpy().astype("float32").tobytes()
+                                    )
+                                else:
+                                    audio_bytes = result.audio.astype("float32").tobytes()
+                                chapter_ffmpeg_proc.stdin.write(audio_bytes)
+                            # Subtitle logic
+                            if self.subtitle_mode != "Disabled":
+                                tokens_list = getattr(result, "tokens", [])
 
-                            # Process every token, regardless of text or timestamps
+                                # Fallback for languages without token support (non-English)
-                            for tok in tokens_list:
+                                # Create a single token representing the entire segment duration
-                                tokens_with_timestamps.append(
+                                if not tokens_list and result.graphemes:
-                                    {
+
-                                        "start": chunk_start + (tok.start_ts or 0),
+                                    class FakeToken:
-                                        "end": chunk_start + (tok.end_ts or 0),
+                                        def __init__(self, text, start, end):
-                                        "text": tok.text,
+                                            self.text = text
-                                        "whitespace": tok.whitespace,
+                                            self.start_ts = start
-                                    }
+                                            self.end_ts = end
-                                )
+                                            self.whitespace = ""
-                                if chapter_out_file or chapter_ffmpeg_proc:
+
-                                    chapter_tokens_with_timestamps.append(
+                                    tokens_list = [
+                                        FakeToken(result.graphemes, 0, chunk_dur)
+                                    ]
+
+                                tokens_with_timestamps = []
+                                chapter_tokens_with_timestamps = []
+
+                                # Process every token, regardless of text or timestamps
+                                for tok in tokens_list:
+                                    tokens_with_timestamps.append(
                                         {
-                                            "start": chapter_current_time
+                                            "start": chunk_start + (tok.start_ts or 0),
-                                            + (tok.start_ts or 0),
+                                            "end": chunk_start + (tok.end_ts or 0),
-                                            "end": chapter_current_time
-                                            + (tok.end_ts or 0),
                                             "text": tok.text,
                                             "whitespace": tok.whitespace,
                                         }
                                     )
-                            # Process tokens according to subtitle mode
+                                    if chapter_out_file or chapter_ffmpeg_proc:
-                            # Global subtitle processing ONLY if merging
+                                        chapter_tokens_with_timestamps.append(
+                                            {
+                                                "start": chapter_current_time
+                                                + (tok.start_ts or 0),
+                                                "end": chapter_current_time
+                                                + (tok.end_ts or 0),
+                                                "text": tok.text,
+                                                "whitespace": tok.whitespace,
+                                            }
+                                        )
+                                # Process tokens according to subtitle mode
+                                # Global subtitle processing ONLY if merging
+                                if merge_chapters_at_end:
+                                    # Incremental subtitle writing for merged output
+                                    new_entries = []
+                                    self._process_subtitle_tokens(
+                                        tokens_with_timestamps,
+                                        new_entries,
+                                        self.max_subtitle_words,
+                                        fallback_end_time=chunk_start + chunk_dur,
+                                    )
+                                    if merged_subtitle_file:
+                                        subtitle_format = getattr(
+                                            self, "subtitle_format", "srt"
+                                        )
+                                        if "ass" in subtitle_format:
+                                            for start, end, text in new_entries:
+                                                start_time = self._ass_time(start)
+                                                end_time = self._ass_time(end)
+                                                # Use karaoke effect for highlighting mode
+                                                effect = (
+                                                    "karaoke"
+                                                    if self.subtitle_mode
+                                                    == "Sentence + Highlighting"
+                                                    else ""
+                                                )
+                                                merged_subtitle_file.write(
+                                                    f"Dialogue: 0,{start_time},{end_time},Default,,{merged_subtitle_margin},{merged_subtitle_margin},0,{effect},{merged_subtitle_alignment_tag}{text}\n"
+                                                )
+                                        else:
+                                            for entry in new_entries:
+                                                start, end, text = entry
+                                                merged_subtitle_file.write(
+                                                    f"{merged_srt_index}\n{self._srt_time(start)} --> {self._srt_time(end)}\n{text}\n\n"
+                                                )
+                                                merged_srt_index += 1
+                                # Per-chapter subtitle processing for both file and ffmpeg_proc
+                                if chapter_out_file or chapter_ffmpeg_proc:
+                                    new_chapter_entries = []
+                                    self._process_subtitle_tokens(
+                                        chapter_tokens_with_timestamps,
+                                        new_chapter_entries,
+                                        self.max_subtitle_words,
+                                        fallback_end_time=chapter_current_time + chunk_dur,
+                                    )
+                                    if chapter_subtitle_file:
+                                        subtitle_format = getattr(
+                                            self, "subtitle_format", "srt"
+                                        )
+                                        if "ass" in subtitle_format:
+                                            for start, end, text in new_chapter_entries:
+                                                start_time = self._ass_time(start)
+                                                end_time = self._ass_time(end)
+                                                # Use karaoke effect for highlighting mode
+                                                effect = (
+                                                    "karaoke"
+                                                    if self.subtitle_mode
+                                                    == "Sentence + Highlighting"
+                                                    else ""
+                                                )
+                                                chapter_subtitle_file.write(
+                                                    f"Dialogue: 0,{start_time},{end_time},Default,,{chapter_subtitle_margin},{chapter_subtitle_margin},0,{effect},{chapter_subtitle_alignment_tag}{text}\n"
+                                                )
+                                        else:
+                                            for entry in new_chapter_entries:
+                                                start, end, text = entry
+                                                chapter_subtitle_file.write(
+                                                    f"{chapter_srt_index}\n{self._srt_time(start)} --> {self._srt_time(end)}\n{text}\n\n"
+                                                )
+                                                chapter_srt_index += 1
                             if merge_chapters_at_end:
-                                # Incremental subtitle writing for merged output
+                                current_time += chunk_dur
-                                new_entries = []
+                                if chapter_out_file or chapter_ffmpeg_proc:
-                                self._process_subtitle_tokens(
+                                    chapter_current_time += chunk_dur
-                                    tokens_with_timestamps,
+                            else:
-                                    new_entries,
+                                if chapter_out_file or chapter_ffmpeg_proc:
-                                    self.max_subtitle_words,
+                                    chapter_current_time += chunk_dur
-                                    fallback_end_time=chunk_start + chunk_dur,
+                            # Calculate percentage based on characters processed
-                                )
+                            percent = min(
-                                if merged_subtitle_file:
+                                int(
-                                    subtitle_format = getattr(
+                                    self.processed_char_count / self.total_char_count * 100
-                                        self, "subtitle_format", "srt"
+                                ),
-                                    )
+                                99,
-                                    if "ass" in subtitle_format:
-                                        for start, end, text in new_entries:
-                                            start_time = self._ass_time(start)
-                                            end_time = self._ass_time(end)
-                                            # Use karaoke effect for highlighting mode
-                                            effect = (
-                                                "karaoke"
-                                                if self.subtitle_mode
-                                                == "Sentence + Highlighting"
-                                                else ""
-                                            )
-                                            merged_subtitle_file.write(
-                                                f"Dialogue: 0,{start_time},{end_time},Default,,{merged_subtitle_margin},{merged_subtitle_margin},0,{effect},{merged_subtitle_alignment_tag}{text}\n"
-                                            )
-                                    else:
-                                        for entry in new_entries:
-                                            start, end, text = entry
-                                            merged_subtitle_file.write(
-                                                f"{merged_srt_index}\n{self._srt_time(start)} --> {self._srt_time(end)}\n{text}\n\n"
-                                            )
-                                            merged_srt_index += 1
-                            # Per-chapter subtitle processing for both file and ffmpeg_proc
-                            if chapter_out_file or chapter_ffmpeg_proc:
-                                new_chapter_entries = []
-                                self._process_subtitle_tokens(
-                                    chapter_tokens_with_timestamps,
-                                    new_chapter_entries,
-                                    self.max_subtitle_words,
-                                    fallback_end_time=chapter_current_time + chunk_dur,
-                                )
-                                if chapter_subtitle_file:
-                                    subtitle_format = getattr(
-                                        self, "subtitle_format", "srt"
-                                    )
-                                    if "ass" in subtitle_format:
-                                        for start, end, text in new_chapter_entries:
-                                            start_time = self._ass_time(start)
-                                            end_time = self._ass_time(end)
-                                            # Use karaoke effect for highlighting mode
-                                            effect = (
-                                                "karaoke"
-                                                if self.subtitle_mode
-                                                == "Sentence + Highlighting"
-                                                else ""
-                                            )
-                                            chapter_subtitle_file.write(
-                                                f"Dialogue: 0,{start_time},{end_time},Default,,{chapter_subtitle_margin},{chapter_subtitle_margin},0,{effect},{chapter_subtitle_alignment_tag}{text}\n"
-                                            )
-                                    else:
-                                        for entry in new_chapter_entries:
-                                            start, end, text = entry
-                                            chapter_subtitle_file.write(
-                                                f"{chapter_srt_index}\n{self._srt_time(start)} --> {self._srt_time(end)}\n{text}\n\n"
-                                            )
-                                            chapter_srt_index += 1
-                        if merge_chapters_at_end:
-                            current_time += chunk_dur
-                            if chapter_out_file or chapter_ffmpeg_proc:
-                                chapter_current_time += chunk_dur
-                        else:
-                            if chapter_out_file or chapter_ffmpeg_proc:
-                                chapter_current_time += chunk_dur
-                        # Calculate percentage based on characters processed
-                        percent = min(
-                            int(
-                                self.processed_char_count / self.total_char_count * 100
-                            ),
-                            99,
-                        )
-
-                        # Calculate ETR based on characters processed
-                        etr_str = "Processing..."
-                        chars_done = self.processed_char_count
-                        elapsed = time.time() - self.etr_start_time
-
-                        # Calculate ETR if enough data is available
-                        if (
-                            chars_done > 0 and elapsed > 0.5
-                        ):  # Check elapsed > 0.5 to avoid instability
-                            avg_time_per_char = elapsed / chars_done
-                            remaining = (
-                                self.total_char_count - self.processed_char_count
                             )
-                            if remaining > 0:
-                                secs = avg_time_per_char * remaining
-                                h = int(secs // 3600)
-                                m = int((secs % 3600) // 60)
-                                s = int(secs % 60)
-                                etr_str = f"{h:02d}:{m:02d}:{s:02d}"
 
-                        # Update progress more frequently (after each result)
+                            # Calculate ETR based on characters processed
-                        self.progress_updated.emit(percent, etr_str)
+                            etr_str = "Processing..."
+                            chars_done = self.processed_char_count
+                            elapsed = time.time() - self.etr_start_time
+
+                            # Calculate ETR if enough data is available
+                            if (
+                                chars_done > 0 and elapsed > 0.5
+                            ):  # Check elapsed > 0.5 to avoid instability
+                                avg_time_per_char = elapsed / chars_done
+                                remaining = (
+                                    self.total_char_count - self.processed_char_count
+                                )
+                                if remaining > 0:
+                                    secs = avg_time_per_char * remaining
+                                    h = int(secs // 3600)
+                                    m = int((secs % 3600) // 60)
+                                    s = int(secs % 60)
+                                    etr_str = f"{h:02d}:{m:02d}:{s:02d}"
+
+                            # Update progress more frequently (after each result)
+                            self.progress_updated.emit(percent, etr_str)
 
                 # Add silence between chapters for merged output (except after the last chapter)
                 if merge_chapters_at_end and chapter_idx < total_chapters:

abogen/pyqt/gui.py

@@ -74,7 +74,7 @@ from abogen.subtitle_utils import (
     calculate_text_length,
 )
 
-from abogen.conversion import ConversionThread, VoicePreviewThread, PlayAudioThread
+from abogen.pyqt.conversion import ConversionThread, VoicePreviewThread, PlayAudioThread, ChapterOptionsDialog, TimestampDetectionDialog
 from abogen.pyqt.book_handler import HandlerDialog
 from abogen.constants import (
     PROGRAM_NAME,
@@ -665,6 +665,11 @@ class TextboxDialog(QDialog):
         self.insert_chapter_btn.clicked.connect(self.insert_chapter_marker)
         button_layout.addWidget(self.insert_chapter_btn)
 
+        self.insert_voice_btn = QPushButton("Insert Voice Marker", self)
+        self.insert_voice_btn.setToolTip("Insert a voice change marker at the cursor position")
+        self.insert_voice_btn.clicked.connect(self.insert_voice_marker)
+        button_layout.addWidget(self.insert_voice_btn)
+
         self.cancel_button = QPushButton("Cancel", self)
         self.cancel_button.clicked.connect(self.reject)
 
@@ -767,6 +772,23 @@ class TextboxDialog(QDialog):
         self.update_char_count()
         self.text_edit.setFocus()
 
+    def insert_voice_marker(self):
+        """Insert a voice marker template at cursor position."""
+        cursor = self.text_edit.textCursor()
+        # Use the currently selected voice as the default
+        try:
+            parent_window = self.parent()
+            if parent_window and hasattr(parent_window, 'selected_voice'):
+                default_voice = parent_window.selected_voice or "af_heart"
+            else:
+                default_voice = "af_heart"
+        except Exception:
+            default_voice = "af_heart"
+        cursor.insertText(f"\n<<VOICE:{default_voice}>>\n")
+        self.text_edit.setTextCursor(cursor)
+        self.update_char_count()
+        self.text_edit.setFocus()
+
 
 def migrate_subtitle_format(config):
     """Convert old subtitle_format values to new internal keys."""
@@ -783,6 +805,108 @@ def migrate_subtitle_format(config):
         save_config(config)
 
 
+class WordSubstitutionsDialog(QDialog):
+    """Dialog for configuring word substitutions and text preprocessing options."""
+
+    def __init__(
+        self,
+        parent=None,
+        initial_list="",
+        initial_case_sensitive=False,
+        initial_caps=False,
+        initial_numerals=False,
+        initial_punctuation=False,
+    ):
+        super().__init__(parent)
+        self.setWindowTitle("Word Substitutions Settings")
+        self.setWindowFlags(
+            Qt.WindowType.Window
+            | Qt.WindowType.WindowCloseButtonHint
+            | Qt.WindowType.WindowMaximizeButtonHint
+        )
+        self.resize(600, 500)
+
+        layout = QVBoxLayout(self)
+
+        # Instructions
+        instructions = QLabel(
+            "Enter word substitutions (one per line) in format: Word|NewWord\n"
+            "  - If nothing after |, the word will be erased completely\n"
+            "  - Substitutions match whole words only (e.g., \"tree\" won't match \"trees\" but will match \"tree's\")\n"
+            "  - By default, matching is case-insensitive (e.g., \"gonna\" matches \"Gonna\", \"GONNA\", etc.)",
+            self,
+        )
+        instructions.setStyleSheet(
+            "padding: 10px; background-color: #f0f0f0; border-radius: 5px;"
+        )
+        instructions.setWordWrap(True)
+        layout.addWidget(instructions)
+
+        # Text edit area
+        self.text_edit = QTextEdit(self)
+        self.text_edit.setAcceptRichText(False)
+        self.text_edit.setPlaceholderText("Word|NewWord")
+        self.text_edit.setPlainText(initial_list)
+        layout.addWidget(self.text_edit)
+
+        # Checkboxes
+        self.case_sensitive_checkbox = QCheckBox(
+            "Case-sensitive word matching", self
+        )
+        self.case_sensitive_checkbox.setChecked(initial_case_sensitive)
+        layout.addWidget(self.case_sensitive_checkbox)
+
+        self.caps_checkbox = QCheckBox("Replace ALL CAPS with lowercase", self)
+        self.caps_checkbox.setChecked(initial_caps)
+        layout.addWidget(self.caps_checkbox)
+
+        self.numerals_checkbox = QCheckBox(
+            "Replace Numerals with Words (e.g., 309 \u2192 three hundred and nine)", self
+        )
+        self.numerals_checkbox.setChecked(initial_numerals)
+        layout.addWidget(self.numerals_checkbox)
+
+        self.punctuation_checkbox = QCheckBox(
+            "Fix Nonstandard Punctuation (curly quotes and other Unicode punctuation that may affect how words sound)",
+            self,
+        )
+        self.punctuation_checkbox.setChecked(initial_punctuation)
+        layout.addWidget(self.punctuation_checkbox)
+
+        # Buttons
+        button_layout = QHBoxLayout()
+        self.cancel_button = QPushButton("Cancel", self)
+        self.cancel_button.clicked.connect(self.reject)
+        self.ok_button = QPushButton("OK", self)
+        self.ok_button.setDefault(True)
+        self.ok_button.clicked.connect(self.accept)
+
+        button_layout.addStretch()
+        button_layout.addWidget(self.cancel_button)
+        button_layout.addWidget(self.ok_button)
+        layout.addLayout(button_layout)
+
+    def get_substitutions_list(self):
+        """Get the substitutions list as plain text."""
+        return self.text_edit.toPlainText()
+
+    def get_case_sensitive(self):
+        """Get whether case-sensitive matching is enabled."""
+        return self.case_sensitive_checkbox.isChecked()
+
+    def get_replace_all_caps(self):
+        """Get whether ALL CAPS replacement is enabled."""
+        return self.caps_checkbox.isChecked()
+
+    def get_replace_numerals(self):
+        """Get whether numeral-to-word conversion is enabled."""
+        return self.numerals_checkbox.isChecked()
+
+    def get_fix_nonstandard_punctuation(self):
+        """Get whether nonstandard punctuation fixing is enabled."""
+        return self.punctuation_checkbox.isChecked()
+
+
 class abogen(QWidget):
     def __init__(self):
         super().__init__()
@@ -833,6 +957,19 @@ class abogen(QWidget):
         self.use_silent_gaps = self.config.get("use_silent_gaps", True)
         self.subtitle_speed_method = self.config.get("subtitle_speed_method", "tts")
         self.use_spacy_segmentation = self.config.get("use_spacy_segmentation", True)
+        # Word substitution settings
+        self.word_substitutions_enabled = self.config.get(
+            "word_substitutions_enabled", False
+        )
+        self.word_substitutions_list = self.config.get("word_substitutions_list", "")
+        self.case_sensitive_substitutions = self.config.get(
+            "case_sensitive_substitutions", False
+        )
+        self.replace_all_caps = self.config.get("replace_all_caps", False)
+        self.replace_numerals = self.config.get("replace_numerals", False)
+        self.fix_nonstandard_punctuation = self.config.get(
+            "fix_nonstandard_punctuation", False
+        )
         self._pending_close_event = None
         self.gpu_ok = False  # Initialize GPU availability status
 
@@ -1071,6 +1208,35 @@ class abogen(QWidget):
         subtitle_layout.addWidget(self.subtitle_combo)
         controls_layout.addLayout(subtitle_layout)
 
+        # Word Substitutions section
+        word_sub_layout = QHBoxLayout()
+        word_sub_layout.setSpacing(7)
+        word_sub_label = QLabel("Word Substitutions:", self)
+        word_sub_layout.addWidget(word_sub_label)
+
+        self.word_sub_combo = QComboBox(self)
+        self.word_sub_combo.addItems(["Disabled", "Enabled"])
+        self.word_sub_combo.setStyleSheet(
+            "QComboBox { min-height: 20px; padding: 6px 12px; }"
+        )
+        self.word_sub_combo.setSizePolicy(
+            QSizePolicy.Policy.Expanding, QSizePolicy.Policy.Fixed
+        )
+        self.word_sub_combo.setCurrentText(
+            "Enabled" if self.word_substitutions_enabled else "Disabled"
+        )
+        self.word_sub_combo.currentTextChanged.connect(self.on_word_sub_changed)
+        word_sub_layout.addWidget(self.word_sub_combo)
+
+        self.btn_word_sub_settings = QPushButton("Settings", self)
+        self.btn_word_sub_settings.setFixedSize(80, 36)
+        self.btn_word_sub_settings.setStyleSheet("QPushButton { padding: 6px 12px; }")
+        self.btn_word_sub_settings.clicked.connect(self.show_word_sub_dialog)
+        self.btn_word_sub_settings.setEnabled(self.word_substitutions_enabled)
+        word_sub_layout.addWidget(self.btn_word_sub_settings)
+
+        controls_layout.addLayout(word_sub_layout)
+
         # Output voice format
         format_layout = QHBoxLayout()
         format_layout.setSpacing(7)
@@ -2015,15 +2181,37 @@ class abogen(QWidget):
                 self.subtitle_speed_method = getattr(
                     queued_item, "subtitle_speed_method", "tts"
                 )
+                # Word substitution settings
+                self.word_substitutions_enabled = getattr(
+                    queued_item, "word_substitutions_enabled", False
+                )
+                self.word_substitutions_list = getattr(
+                    queued_item, "word_substitutions_list", ""
+                )
+                self.case_sensitive_substitutions = getattr(
+                    queued_item, "case_sensitive_substitutions", False
+                )
+                self.replace_all_caps = getattr(queued_item, "replace_all_caps", False)
+                self.replace_numerals = getattr(queued_item, "replace_numerals", False)
+                self.fix_nonstandard_punctuation = getattr(
+                    queued_item, "fix_nonstandard_punctuation", False
+                )
 
-                # This ensures that if conversion.py (or utils) reads from config/disk 
+                # This ensures that if conversion.py (or utils) reads from config/disk
                 # instead of using passed arguments, it sees the correct queue values.
                 self.config["replace_single_newlines"] = self.replace_single_newlines
                 self.config["subtitle_mode"] = self.subtitle_mode
                 self.config["selected_format"] = self.selected_format
                 self.config["use_silent_gaps"] = self.use_silent_gaps
                 self.config["subtitle_speed_method"] = self.subtitle_speed_method
-                
+                # Word substitution settings
+                self.config["word_substitutions_enabled"] = self.word_substitutions_enabled
+                self.config["word_substitutions_list"] = self.word_substitutions_list
+                self.config["case_sensitive_substitutions"] = self.case_sensitive_substitutions
+                self.config["replace_all_caps"] = self.replace_all_caps
+                self.config["replace_numerals"] = self.replace_numerals
+                self.config["fix_nonstandard_punctuation"] = self.fix_nonstandard_punctuation
+
                 # Sync Voice/Profile in config
                 self.config["selected_voice"] = self.selected_voice
                 if "selected_profile_name" in self.config:
@@ -2179,6 +2367,21 @@ class abogen(QWidget):
             self.conversion_thread.subtitle_speed_method = self.subtitle_speed_method
             # Pass use_spacy_segmentation setting
             self.conversion_thread.use_spacy_segmentation = self.use_spacy_segmentation
+            # Pass word substitution settings
+            self.conversion_thread.word_substitutions_enabled = (
+                self.word_substitutions_enabled
+            )
+            self.conversion_thread.word_substitutions_list = (
+                self.word_substitutions_list
+            )
+            self.conversion_thread.case_sensitive_substitutions = (
+                self.case_sensitive_substitutions
+            )
+            self.conversion_thread.replace_all_caps = self.replace_all_caps
+            self.conversion_thread.replace_numerals = self.replace_numerals
+            self.conversion_thread.fix_nonstandard_punctuation = (
+                self.fix_nonstandard_punctuation
+            )
             # Pass separate_chapters_format setting
             self.conversion_thread.separate_chapters_format = (
                 self.separate_chapters_format
@@ -2927,6 +3130,41 @@ class abogen(QWidget):
         self.config["use_gpu"] = self.use_gpu
         save_config(self.config)
 
+    def on_word_sub_changed(self, text):
+        """Handle word substitution dropdown change."""
+        self.word_substitutions_enabled = text == "Enabled"
+        self.btn_word_sub_settings.setEnabled(self.word_substitutions_enabled)
+
+        # Save to config
+        self.config["word_substitutions_enabled"] = self.word_substitutions_enabled
+        save_config(self.config)
+
+    def show_word_sub_dialog(self):
+        """Show word substitutions settings dialog."""
+        dialog = WordSubstitutionsDialog(
+            self,
+            initial_list=self.word_substitutions_list,
+            initial_case_sensitive=self.case_sensitive_substitutions,
+            initial_caps=self.replace_all_caps,
+            initial_numerals=self.replace_numerals,
+            initial_punctuation=self.fix_nonstandard_punctuation,
+        )
+
+        if dialog.exec() == QDialog.DialogCode.Accepted:
+            self.word_substitutions_list = dialog.get_substitutions_list()
+            self.case_sensitive_substitutions = dialog.get_case_sensitive()
+            self.replace_all_caps = dialog.get_replace_all_caps()
+            self.replace_numerals = dialog.get_replace_numerals()
+            self.fix_nonstandard_punctuation = dialog.get_fix_nonstandard_punctuation()
+
+            # Save all settings to config
+            self.config["word_substitutions_list"] = self.word_substitutions_list
+            self.config["case_sensitive_substitutions"] = self.case_sensitive_substitutions
+            self.config["replace_all_caps"] = self.replace_all_caps
+            self.config["replace_numerals"] = self.replace_numerals
+            self.config["fix_nonstandard_punctuation"] = self.fix_nonstandard_punctuation
+            save_config(self.config)
+
     def cleanup_conversion_thread(self):
         # Stop conversion thread
         if (
@@ -2991,8 +3229,6 @@ class abogen(QWidget):
         """Show dialog to ask user about chapter processing options when chapters are detected in a .txt file"""
         # Check if this is a timestamp detection (-1) or chapter detection
         if chapter_count == -1:
-            from abogen.conversion import TimestampDetectionDialog
-
             dialog = TimestampDetectionDialog(parent=self)
             dialog.setWindowModality(Qt.WindowModality.ApplicationModal)
 
@@ -3007,8 +3243,6 @@ class abogen(QWidget):
             return
 
         # Normal chapter detection
-        from abogen.conversion import ChapterOptionsDialog
-
         dialog = ChapterOptionsDialog(chapter_count, parent=self)
         dialog.setWindowModality(Qt.WindowModality.ApplicationModal)
 

abogen/pyqt/queue_manager_gui.py

@@ -35,6 +35,12 @@ OVERRIDE_FIELDS = [
     "replace_single_newlines",
     "use_silent_gaps",
     "subtitle_speed_method",
+    "word_substitutions_enabled",
+    "word_substitutions_list",
+    "case_sensitive_substitutions",
+    "replace_all_caps",
+    "replace_numerals",
+    "fix_nonstandard_punctuation",
 ]
 
 
@@ -474,6 +480,21 @@ class QueueManager(QDialog):
             attrs["subtitle_speed_method"] = getattr(
                 parent, "subtitle_speed_method", "tts"
             )
+            # word substitutions
+            attrs["word_substitutions_enabled"] = getattr(
+                parent, "word_substitutions_enabled", False
+            )
+            attrs["word_substitutions_list"] = getattr(
+                parent, "word_substitutions_list", ""
+            )
+            attrs["case_sensitive_substitutions"] = getattr(
+                parent, "case_sensitive_substitutions", False
+            )
+            attrs["replace_all_caps"] = getattr(parent, "replace_all_caps", False)
+            attrs["replace_numerals"] = getattr(parent, "replace_numerals", False)
+            attrs["fix_nonstandard_punctuation"] = getattr(
+                parent, "fix_nonstandard_punctuation", False
+            )
             # book handler options
             attrs["save_chapters_separately"] = getattr(
                 parent, "save_chapters_separately", None

abogen/pyqt/queued_item.py

@@ -19,3 +19,10 @@ class QueuedItem:
     save_base_path: str = None
     save_chapters_separately: bool = None
     merge_chapters_at_end: bool = None
+    # Word Substitution fields
+    word_substitutions_enabled: bool = False
+    word_substitutions_list: str = ""
+    case_sensitive_substitutions: bool = False
+    replace_all_caps: bool = False
+    replace_numerals: bool = False
+    fix_nonstandard_punctuation: bool = False

abogen/subtitle_utils.py

@@ -15,6 +15,8 @@ _ASS_STYLING_PATTERN = re.compile(r"\{[^}]+\}")
 _ASS_NEWLINE_N_PATTERN = re.compile(r"\\N")
 _ASS_NEWLINE_LOWER_N_PATTERN = re.compile(r"\\n")
 _CHAPTER_MARKER_SEARCH_PATTERN = re.compile(r"<<CHAPTER_MARKER:(.*?)>>")
+_VOICE_MARKER_PATTERN = re.compile(r"<<VOICE:[^>]*>>")
+_VOICE_MARKER_SEARCH_PATTERN = re.compile(r"<<VOICE:(.*?)>>")
 _WEBVTT_HEADER_PATTERN = re.compile(r"^WEBVTT.*?\n", re.MULTILINE)
 _VTT_STYLE_PATTERN = re.compile(r"STYLE\s*\n.*?(?=\n\n|$)", re.DOTALL)
 _VTT_NOTE_PATTERN = re.compile(r"NOTE\s*\n.*?(?=\n\n|$)", re.DOTALL)
@@ -31,17 +33,19 @@ _LINUX_ILLEGAL_CHARS_PATTERN = re.compile(r"[/\x00]")
 
 
 def clean_subtitle_text(text):
-    """Remove chapter markers and metadata tags from subtitle text."""
+    """Remove chapter markers, voice markers, and metadata tags from subtitle text."""
     # Use pre-compiled patterns for better performance
     text = _METADATA_TAG_PATTERN.sub("", text)
     text = _CHAPTER_MARKER_PATTERN.sub("", text)
+    text = _VOICE_MARKER_PATTERN.sub("", text)
     return text.strip()
 
 
 def calculate_text_length(text):
     # Use pre-compiled patterns for better performance
-    # Ignore chapter markers and metadata patterns in a single pass
+    # Ignore chapter markers, voice markers, and metadata patterns in a single pass
     text = _CHAPTER_MARKER_PATTERN.sub("", text)
+    text = _VOICE_MARKER_PATTERN.sub("", text)
     text = _METADATA_TAG_PATTERN.sub("", text)
     # Ignore newlines and leading/trailing spaces
     text = text.replace("\n", "").strip()
@@ -459,3 +463,122 @@ def sanitize_name_for_os(name, is_folder=True):
         sanitized = sanitized[:255].rstrip(". ")
 
     return sanitized
+
+
+def validate_voice_name(voice_name):
+    """Validate voice name against VOICES_INTERNAL list (case-insensitive).
+    Handles both single voices and formulas like 'af_heart*0.5 + am_echo*0.5'.
+
+    Args:
+        voice_name: Voice name or formula string to validate
+
+    Returns:
+        Tuple of (is_valid, invalid_voice_name):
+            - is_valid: True if all voices in the name/formula are valid
+            - invalid_voice_name: The first invalid voice found, or None if all valid
+    """
+    from abogen.constants import VOICES_INTERNAL
+
+    # Create case-insensitive lookup set (done once per call)
+    voice_lookup_lower = {v.lower() for v in VOICES_INTERNAL}
+    voice_name = voice_name.strip()
+
+    # Check if it's a formula (contains *)
+    if "*" in voice_name:
+        # Extract voice names from formula
+        voices = voice_name.split("+")
+        for term in voices:
+            if "*" in term:
+                base_voice = term.split("*")[0].strip()
+                # Case-insensitive comparison
+                if base_voice.lower() not in voice_lookup_lower:
+                    return False, base_voice
+        return True, None
+    else:
+        # Single voice - case-insensitive comparison
+        if voice_name.lower() not in voice_lookup_lower:
+            return False, voice_name
+        return True, None
+
+
+def split_text_by_voice_markers(text, default_voice):
+    """Split text by voice markers, returning list of (voice, text) tuples.
+
+    IMPORTANT: Returns the last voice used so it can persist across chapters.
+    Voice names are normalized to lowercase to match VOICES_INTERNAL.
+
+    Args:
+        text: Text potentially containing <<VOICE:name>> markers
+        default_voice: Voice to use if no markers found or before first marker
+
+    Returns:
+        Tuple of (segments_list, last_voice_used, valid_count, invalid_count):
+            - segments_list: List of (voice_name, segment_text) tuples
+            - last_voice_used: The voice that should continue into next chapter
+            - valid_count: Number of valid voice markers processed
+            - invalid_count: Number of invalid voice markers skipped
+    """
+    from abogen.constants import VOICES_INTERNAL
+
+    voice_splits = list(_VOICE_MARKER_SEARCH_PATTERN.finditer(text))
+
+    if not voice_splits:
+        # No voice markers, return entire text with default voice
+        return [(default_voice, text)], default_voice, 0, 0
+
+    segments = []
+    current_voice = default_voice
+    valid_markers = 0
+    invalid_markers = 0
+
+    # Text before first marker uses default voice
+    first_start = voice_splits[0].start()
+    if first_start > 0:
+        intro_text = text[:first_start].strip()
+        if intro_text:
+            segments.append((current_voice, intro_text))
+
+    # Process each voice marker
+    for idx, match in enumerate(voice_splits):
+        voice_name = match.group(1).strip()
+        start = match.end()
+        end = voice_splits[idx + 1].start() if idx + 1 < len(voice_splits) else len(text)
+        segment_text = text[start:end].strip()
+
+        # Validate voice name
+        is_valid, invalid_voice = validate_voice_name(voice_name)
+        if is_valid:
+            # Normalize to lowercase to match canonical form
+            # Handle both single voices and formulas
+            if "*" in voice_name:
+                # Normalize each voice in the formula
+                normalized_parts = []
+                for part in voice_name.split("+"):
+                    part = part.strip()
+                    if "*" in part:
+                        voice_part, weight = part.split("*", 1)
+                        # Find the canonical (lowercase) voice name
+                        voice_part_lower = voice_part.strip().lower()
+                        canonical_voice = next(
+                            (v for v in VOICES_INTERNAL if v.lower() == voice_part_lower),
+                            voice_part.strip()
+                        )
+                        normalized_parts.append(f"{canonical_voice}*{weight.strip()}")
+                current_voice = " + ".join(normalized_parts)
+            else:
+                # Find the canonical (lowercase) voice name
+                voice_name_lower = voice_name.lower()
+                current_voice = next(
+                    (v for v in VOICES_INTERNAL if v.lower() == voice_name_lower),
+                    voice_name
+                )
+            valid_markers += 1
+        else:
+            # Invalid voice - stay with previous voice
+            invalid_markers += 1
+
+        if segment_text:
+            segments.append((current_voice, segment_text))
+
+    # Return segments, last voice, and counts
+    return segments, current_voice, valid_markers, invalid_markers

abogen/text_extractor.py

@@ -1023,8 +1023,13 @@ class EpubExtractor:
         if not html:
             return ""
         soup = BeautifulSoup(html, "html.parser")
-        for tag in soup.find_all(["p", "div"]):
+
+        # Add line breaks after block-level elements to ensure pauses in speech
+        for tag in soup.find_all(
+            ["p", "div", "h1", "h2", "h3", "h4", "h5", "h6", "li", "blockquote"]
+        ):
             tag.append("\n\n")
+
         for ol in soup.find_all("ol"):
             start_attr = ol.get("start")
             try:

abogen/webui/routes/utils/preview.py

@@ -17,10 +17,43 @@ _preview_pipeline_lock = threading.Lock()
 def _select_device() -> str:
     import platform
 
+    try:
+        import torch  # type: ignore[import-not-found]
+    except Exception:
+        return "cpu"
+
     system = platform.system()
     if system == "Darwin" and platform.processor() == "arm":
-        return "mps"
+        try:
-    return "cuda"
+            if torch.backends.mps.is_available():
+                return "mps"
+        except Exception:
+            pass
+        return "cpu"
+
+    try:
+        if torch.cuda.is_available():
+            return "cuda"
+    except Exception:
+        pass
+    return "cpu"
+
+
+def _resolve_pipeline(language: str, use_gpu: bool) -> Tuple[Any, bool]:
+    devices: List[str] = ["cpu"]
+    if use_gpu:
+        preferred = _select_device()
+        if preferred != "cpu":
+            devices.insert(0, preferred)
+
+    last_error: Optional[Exception] = None
+    for device in devices:
+        try:
+            return get_preview_pipeline(language, device), device != "cpu"
+        except Exception as exc:
+            last_error = exc
+
+    raise RuntimeError("Preview pipeline is unavailable") from last_error
 
 
 def _to_float32(audio_segment) -> np.ndarray:
@@ -115,15 +148,7 @@ def generate_preview_audio(
             total_steps=supertonic_total_steps,
         )
     else:
-        device = "cpu"
+        pipeline, pipeline_uses_gpu = _resolve_pipeline(language, use_gpu)
-        if use_gpu:
-            try:
-                device = _select_device()
-            except Exception:
-                device = "cpu"
-                use_gpu = False
-
-        pipeline = get_preview_pipeline(language, device)
         if pipeline is None:
             raise RuntimeError("Preview pipeline is unavailable")
 
@@ -131,7 +156,7 @@ def generate_preview_audio(
         if voice_spec and "*" in voice_spec:
             from abogen.voice_formulas import get_new_voice
 
-            voice_choice = get_new_voice(pipeline, voice_spec, use_gpu)
+            voice_choice = get_new_voice(pipeline, voice_spec, pipeline_uses_gpu)
 
         segments = pipeline(
             normalized_text,

abogen/word_substitution.py

@@ -0,0 +1,254 @@
+"""
+Word substitution module for text-to-speech preprocessing.
+
+This module provides functionality to:
+- Replace words/phrases with custom text
+- Convert ALL CAPS to lowercase
+- Convert numerals to words
+- Fix nonstandard punctuation for TTS compatibility
+
+All substitutions preserve special markers (chapter, voice, metadata, timestamps).
+"""
+
+import re
+
+from abogen.subtitle_utils import (
+    _CHAPTER_MARKER_PATTERN,
+    _VOICE_MARKER_PATTERN,
+    _METADATA_TAG_PATTERN,
+    _TIMESTAMP_ONLY_PATTERN,
+)
+
+
+def apply_word_substitutions(
+    text,
+    substitutions_list_str,
+    case_sensitive=False,
+    replace_all_caps=False,
+    replace_numerals=False,
+    fix_nonstandard_punctuation=False,
+):
+    """
+    Apply word substitutions to text while preserving markers.
+
+    Args:
+        text: Input text
+        substitutions_list_str: Newline-separated "Word|NewWord" pairs
+        case_sensitive: If True, match words case-sensitively
+        replace_all_caps: Convert ALL CAPS words to lowercase
+        replace_numerals: Convert numbers to words
+        fix_nonstandard_punctuation: Fix curly quotes, em/en dashes, etc.
+
+    Returns:
+        Modified text
+    """
+    # Apply nonstandard punctuation fixes FIRST (if enabled)
+    if fix_nonstandard_punctuation:
+        text = fix_punctuation(text)
+
+    # Parse substitutions list
+    substitutions = parse_substitutions_list(substitutions_list_str)
+
+    # Split text into segments (markers vs content)
+    segments = split_text_preserving_markers(text)
+
+    # Process each segment
+    processed_segments = []
+    for segment_type, segment_text in segments:
+        if segment_type == "marker":
+            # Preserve markers unchanged
+            processed_segments.append(segment_text)
+        else:
+            # Apply substitutions to content
+            processed_text = segment_text
+
+            # Apply word substitutions
+            if substitutions:
+                processed_text = apply_word_replacements(
+                    processed_text, substitutions, case_sensitive
+                )
+
+            # Apply ALL CAPS conversion
+            if replace_all_caps:
+                processed_text = convert_all_caps_to_lowercase(processed_text)
+
+            # Apply numeral conversion
+            if replace_numerals:
+                processed_text = convert_numerals_to_words(processed_text)
+
+            processed_segments.append(processed_text)
+
+    return "".join(processed_segments)
+
+
+def parse_substitutions_list(substitutions_str):
+    """
+    Parse newline-separated "Word|NewWord" format.
+
+    Args:
+        substitutions_str: String with substitutions, one per line
+
+    Returns:
+        List of tuples: [(word, replacement), ...]
+    """
+    substitutions = []
+    for line in substitutions_str.strip().split("\n"):
+        line = line.strip()
+        if not line or "|" not in line:
+            continue
+
+        parts = line.split("|", 1)
+        if len(parts) == 2:
+            word = parts[0].strip()
+            replacement = parts[1].strip()
+            if word:  # Only add if word is not empty
+                substitutions.append((word, replacement))
+
+    return substitutions
+
+
+def split_text_preserving_markers(text):
+    """
+    Split text into segments alternating between markers and content.
+
+    Args:
+        text: Input text with potential markers
+
+    Returns:
+        List of tuples: [("marker"|"content", text), ...]
+    """
+    # Combined pattern for all markers and timestamps
+    marker_pattern = re.compile(
+        r"(<<CHAPTER_MARKER:[^>]*>>|<<VOICE:[^>]*>>|<<METADATA_[^:]+:[^>]*>>|\d{1,2}:\d{2}:\d{2}(?:[.,]\d{1,3})?)"
+    )
+
+    segments = []
+    last_end = 0
+
+    for match in marker_pattern.finditer(text):
+        # Content before marker
+        if match.start() > last_end:
+            segments.append(("content", text[last_end : match.start()]))
+
+        # Marker itself
+        segments.append(("marker", match.group(0)))
+        last_end = match.end()
+
+    # Remaining content after last marker
+    if last_end < len(text):
+        segments.append(("content", text[last_end:]))
+
+    return segments
+
+
+def apply_word_replacements(text, substitutions, case_sensitive=False):
+    """
+    Apply word substitutions using whole-word matching.
+
+    Args:
+        text: Input text
+        substitutions: List of (word, replacement) tuples
+        case_sensitive: If True, match case-sensitively
+
+    Returns:
+        Text with substitutions applied
+    """
+    for word, replacement in substitutions:
+        # Use word boundaries for exact matching
+        # Escape special regex characters
+        escaped_word = re.escape(word)
+        pattern = re.compile(
+            r"\b" + escaped_word + r"\b",
+            0 if case_sensitive else re.IGNORECASE,
+        )
+        text = pattern.sub(replacement, text)
+
+    return text
+
+
+def convert_all_caps_to_lowercase(text):
+    """
+    Convert ALL CAPS words to lowercase.
+
+    Args:
+        text: Input text
+
+    Returns:
+        Text with ALL CAPS converted to lowercase
+    """
+
+    def replace_caps(match):
+        word = match.group(0)
+        # Convert to lowercase
+        return word.lower()
+
+    # Match words that are ALL CAPS (2+ letters)
+    pattern = re.compile(r"\b[A-Z]{2,}\b")
+    return pattern.sub(replace_caps, text)
+
+
+def convert_numerals_to_words(text):
+    """
+    Convert numerals to words using num2words library.
+
+    Args:
+        text: Input text
+
+    Returns:
+        Text with numerals converted to words
+    """
+    try:
+        from num2words import num2words
+    except ImportError:
+        # If num2words not available, return unchanged
+        return text
+
+    def replace_number(match):
+        try:
+            number = int(match.group(0))
+            # Convert to words in English
+            return num2words(number)
+        except Exception:
+            # If conversion fails, return original
+            return match.group(0)
+
+    # Match integers (but not timestamps or other patterns)
+    # Negative lookbehind/ahead to avoid timestamps
+    pattern = re.compile(r"(?<!\d:)\b\d+\b(?!:\d)")
+    return pattern.sub(replace_number, text)
+
+
+def fix_punctuation(text):
+    """
+    Convert nonstandard punctuation to standard equivalents.
+
+    This helps TTS engines pronounce words correctly by converting:
+    - Curly quotes to straight quotes
+    - Ellipsis to three periods
+
+    Args:
+        text: Input text
+
+    Returns:
+        Text with nonstandard punctuation fixed
+    """
+    # Define replacements
+    replacements = {
+        # Curly double quotes
+        "\u201c": '"',  # Left double quotation mark
+        "\u201d": '"',  # Right double quotation mark
+        "\u201e": '"',  # Double low-9 quotation mark
+        # Curly single quotes
+        "\u2018": "'",  # Left single quotation mark
+        "\u2019": "'",  # Right single quotation mark
+        "\u201a": "'",  # Single low-9 quotation mark
+        "\u201b": "'",  # Single high-reversed-9 quotation mark
+        # Other punctuation
+        "\u2026": "...",  # Ellipsis
+    }
+
+    # Apply all replacements
+    for old_char, new_char in replacements.items():
+        text = text.replace(old_char, new_char)
+
+    return text

pyproject.toml

@@ -50,6 +50,8 @@ dependencies = [
   "num2words>=0.5.13",
   "httpx>=0.27.0",
   "PyQt6>=6.5.0",
+  "flet>=0.85.1",
+  "msgpack>=1.0.0",
 ]
 
 classifiers = [
@@ -77,11 +79,12 @@ allow-direct-references = true
 
 
 [project.gui-scripts]
-abogen = "abogen.pyqt.main:main"
+abogen = "abogen.frontend.main:main"
 
 [project.scripts]
+abogen-ui = "abogen.frontend.main:main"
+abogen-web = "abogen.frontend.main:main_web"
 abogen-cli = "abogen.webui.app:main"
-abogen-web = "abogen.webui.app:main"
 abogen-pyqt = "abogen.pyqt.main:main"
 
 [tool.hatch.build.targets.sdist]
@@ -96,8 +99,6 @@ exclude = [
 [tool.hatch.build.targets.wheel]
 packages = ["abogen"]
 
-[tool.hatch.build]
-include = ["abogen/webui/templates/**", "abogen/webui/static/**"]
 
 [tool.hatch.version]
 path = "abogen/VERSION"