feat: Add chapter intro delay setting and implement in conversion process

abogen/web/conversion_runner.py

@@ -204,6 +204,18 @@ def _escape_ffmetadata_value(value: str) -> str:
     return escaped
 
 
+def _metadata_to_ffmpeg_args(metadata: Dict[str, Any]) -> List[str]:
+    args: List[str] = []
+    for key, value in (metadata or {}).items():
+        if value in (None, ""):
+            continue
+        key_str = str(key).strip()
+        if not key_str:
+            continue
+        args.extend(["-metadata", f"{key_str}={value}"])
+    return args
+
+
 def _render_ffmetadata(metadata: Dict[str, Any], chapters: List[Dict[str, Any]]) -> str:
     lines: List[str] = [";FFMETADATA1"]
     for key, value in (metadata or {}).items():
@@ -274,10 +286,12 @@ def _embed_m4b_metadata(
         if candidate.exists():
             cover_path = candidate
 
-    if not ffmetadata_path and not cover_path:
+    metadata_args = _metadata_to_ffmpeg_args(metadata_map)
+
+    if not ffmetadata_path and not cover_path and not metadata_args:
         return
 
-    job.add_log("Embedding chapters and cover metadata into m4b output")
+    job.add_log("Embedding metadata into m4b output")
 
     command: List[str] = ["ffmpeg", "-y", "-i", str(audio_path)]
     metadata_index: Optional[int] = None
@@ -311,6 +325,9 @@ def _embed_m4b_metadata(
     else:
         command += ["-map_metadata", "0"]
 
+    if metadata_args:
+        command.extend(metadata_args)
+
     command += ["-movflags", "+faststart+use_metadata_tags"]
 
     temp_output = audio_path.with_suffix(audio_path.suffix + ".tmp")
@@ -332,6 +349,7 @@ def _embed_m4b_metadata(
         raise RuntimeError(f"ffmpeg failed to embed metadata (exit code {return_code})")
 
     temp_output.replace(audio_path)
+    job.add_log("Embedded metadata and chapters into m4b output", level="info")
 
 
 def run_conversion_job(job: Job) -> None:
@@ -408,6 +426,83 @@ def run_conversion_job(job: Job) -> None:
         total_chapters = len(extraction.chapters)
         job.add_log(f"Detected {total_chapters} chapter{'s' if total_chapters != 1 else ''}")
 
+        def emit_text(
+            text: str,
+            *,
+            voice_choice: Any,
+            chapter_sink: Optional[AudioSink],
+            preview_prefix: Optional[str] = None,
+            split_pattern: Optional[str] = SPLIT_PATTERN,
+        ) -> int:
+            nonlocal processed_chars, subtitle_index, current_time
+            normalized = _normalize_for_pipeline(text)
+            local_segments = 0
+
+            for segment in pipeline(
+                normalized,
+                voice=voice_choice,
+                speed=job.speed,
+                split_pattern=split_pattern,
+            ):
+                canceller()
+                graphemes_raw = getattr(segment, "graphemes", "") or ""
+                graphemes = graphemes_raw.strip()
+
+                audio = _to_float32(getattr(segment, "audio", None))
+                if audio.size == 0:
+                    continue
+
+                local_segments += 1
+                if chapter_sink:
+                    chapter_sink.write(audio)
+                if audio_sink:
+                    audio_sink.write(audio)
+
+                duration = len(audio) / SAMPLE_RATE
+                processed_chars += len(graphemes)
+                job.processed_characters = processed_chars
+                if job.total_characters:
+                    job.progress = min(processed_chars / job.total_characters, 0.999)
+                else:
+                    job.progress = 0.0 if processed_chars == 0 else 0.999
+
+                preview_text = graphemes or (graphemes_raw[:80] if graphemes_raw else "[silence]")
+                prefix = f"{preview_prefix} · " if preview_prefix else ""
+                job.add_log(f"{prefix}{processed_chars:,}/{job.total_characters or '—'}: {preview_text[:80]}")
+
+                if subtitle_writer and audio_sink and graphemes:
+                    subtitle_writer.write_segment(
+                        index=subtitle_index,
+                        text=graphemes,
+                        start=current_time,
+                        end=current_time + duration,
+                    )
+                    subtitle_index += 1
+
+                if audio_sink:
+                    current_time += duration
+
+            return local_segments
+
+        def append_silence(
+            duration_seconds: float,
+            *,
+            include_in_chapter: bool,
+            chapter_sink: Optional[AudioSink],
+        ) -> None:
+            nonlocal current_time
+            if duration_seconds <= 0:
+                return
+            samples = int(round(duration_seconds * SAMPLE_RATE))
+            if samples <= 0:
+                return
+            silence = np.zeros(samples, dtype="float32")
+            if include_in_chapter and chapter_sink:
+                chapter_sink.write(silence)
+            if audio_sink:
+                audio_sink.write(silence)
+                current_time += duration_seconds
+
         for idx, chapter in enumerate(extraction.chapters, start=1):
             canceller()
             job.add_log(f"Processing chapter {idx}/{total_chapters}: {chapter.title}")
@@ -425,71 +520,54 @@ def run_conversion_job(job: Job) -> None:
                 voice_choice = _resolve_voice(pipeline, chapter_voice_spec, job.use_gpu)
                 voice_cache[chapter_voice_spec] = voice_choice
 
-            chapter_sink_stack = ExitStack()
-            chapter_sink: Optional[AudioSink] = None
             chapter_audio_path: Optional[Path] = None
-
-            if chapter_dir is not None:
-                chapter_audio_path = _build_output_path(
-                    chapter_dir,
-                    f"{Path(job.original_filename).stem}_{_slugify(chapter.title, idx)}",
-                    job.separate_chapters_format,
-                )
-                chapter_sink = _open_audio_sink(
-                    chapter_audio_path,
-                    job,
-                    chapter_sink_stack,
-                    fmt=job.separate_chapters_format,
-                )
-
             segments_emitted = 0
-            tts_input = _normalize_for_pipeline(chapter.text)
 
-            for segment in pipeline(
+            with ExitStack() as chapter_sink_stack:
-                tts_input,
+                chapter_sink: Optional[AudioSink] = None
-                voice=voice_choice,
-                speed=job.speed,
-                split_pattern=SPLIT_PATTERN,
-            ):
-                canceller()
-                graphemes_raw = getattr(segment, "graphemes", "") or ""
-                graphemes = graphemes_raw.strip()
 
-                audio = _to_float32(getattr(segment, "audio", None))
+                if chapter_dir is not None:
-                if audio.size == 0:
+                    chapter_audio_path = _build_output_path(
-                    continue
+                        chapter_dir,
-
+                        f"{Path(job.original_filename).stem}_{_slugify(chapter.title, idx)}",
-                segments_emitted += 1
+                        job.separate_chapters_format,
-                if chapter_sink:
+                    )
-                    chapter_sink.write(audio)
+                    chapter_sink = _open_audio_sink(
-                if audio_sink:
+                        chapter_audio_path,
-                    audio_sink.write(audio)
+                        job,
-
+                        chapter_sink_stack,
-                duration = len(audio) / SAMPLE_RATE
+                        fmt=job.separate_chapters_format,
-                processed_chars += len(graphemes)
-                job.processed_characters = processed_chars
-                if job.total_characters:
-                    job.progress = min(processed_chars / job.total_characters, 0.999)
-                else:
-                    job.progress = 0.0 if processed_chars == 0 else 0.999
-
-                preview_text = graphemes or (graphemes_raw[:80] if graphemes_raw else "[silence]")
-                job.add_log(f"{processed_chars:,}/{job.total_characters or '—'}: {preview_text[:80]}")
-
-                if subtitle_writer and audio_sink and graphemes:
-                    subtitle_writer.write_segment(
-                        index=subtitle_index,
-                        text=graphemes,
-                        start=current_time,
-                        end=current_time + duration,
                     )
-                    subtitle_index += 1
 
-                if audio_sink:
+                speak_heading = bool(chapter.title.strip())
-                    current_time += duration
+                if speak_heading:
+                    stripped_title = chapter.title.strip()
+                    if stripped_title:
+                        first_line = next((line.strip() for line in chapter.text.splitlines() if line.strip()), "")
+                        if first_line and first_line.casefold() == stripped_title.casefold():
+                            speak_heading = False
 
-            if chapter_sink:
+                if speak_heading:
-                chapter_sink_stack.close()
+                    heading_segments = emit_text(
+                        chapter.title,
+                        voice_choice=voice_choice,
+                        chapter_sink=chapter_sink,
+                        preview_prefix=f"Chapter {idx} title",
+                        split_pattern=SPLIT_PATTERN,
+                    )
+                    segments_emitted += heading_segments
+                    if heading_segments > 0 and job.chapter_intro_delay > 0:
+                        append_silence(
+                            job.chapter_intro_delay,
+                            include_in_chapter=True,
+                            chapter_sink=chapter_sink,
+                        )
+
+                segments_emitted += emit_text(
+                    chapter.text,
+                    voice_choice=voice_choice,
+                    chapter_sink=chapter_sink,
+                )
 
             chapter_end_time = current_time
 
@@ -511,12 +589,12 @@ def run_conversion_job(job: Job) -> None:
                 and idx < total_chapters
                 and job.silence_between_chapters > 0
             ):
-                silence_samples = int(job.silence_between_chapters * SAMPLE_RATE)
+                append_silence(
-                if silence_samples > 0:
+                    job.silence_between_chapters,
-                    silence = np.zeros(silence_samples, dtype="float32")
+                    include_in_chapter=False,
-                    audio_sink.write(silence)
+                    chapter_sink=None,
-                    current_time += job.silence_between_chapters
+                )
-                    chapter_end_time = current_time
+                chapter_end_time = current_time
 
             chapter_markers.append(
                 {
@@ -730,9 +808,7 @@ def _build_ffmpeg_command(path: Path, fmt: str, metadata: Optional[Dict[str, str
         base += ["-c:a", "copy"]
 
     if metadata:
-        for key, value in metadata.items():
+        base.extend(_metadata_to_ffmpeg_args(metadata))
-            if value:
-                base += ["-metadata", f"{key}={value}"]
     base.append(str(path))
     return base
 

abogen/web/routes.py

@@ -135,7 +135,7 @@ BOOLEAN_SETTINGS = {
     "save_as_project",
 }
 
-FLOAT_SETTINGS = {"silence_between_chapters"}
+FLOAT_SETTINGS = {"silence_between_chapters", "chapter_intro_delay"}
 INT_SETTINGS = {"max_subtitle_words"}
 
 
@@ -156,6 +156,7 @@ def _settings_defaults() -> Dict[str, Any]:
         "save_as_project": False,
         "separate_chapters_format": "wav",
         "silence_between_chapters": 2.0,
+        "chapter_intro_delay": 0.5,
         "max_subtitle_words": 50,
     }
 
@@ -420,6 +421,9 @@ def settings_page() -> Response | str:
         updated["silence_between_chapters"] = _coerce_float(
             form.get("silence_between_chapters"), defaults["silence_between_chapters"]
         )
+        updated["chapter_intro_delay"] = _coerce_float(
+            form.get("chapter_intro_delay"), defaults["chapter_intro_delay"]
+        )
         updated["max_subtitle_words"] = _coerce_int(
             form.get("max_subtitle_words"), defaults["max_subtitle_words"]
         )
@@ -801,6 +805,7 @@ def enqueue_job() -> Response:
     save_as_project = settings["save_as_project"]
     separate_chapters_format = settings["separate_chapters_format"]
     silence_between_chapters = settings["silence_between_chapters"]
+    chapter_intro_delay = settings["chapter_intro_delay"]
     max_subtitle_words = settings["max_subtitle_words"]
 
     pending = PendingJob(
@@ -830,6 +835,7 @@ def enqueue_job() -> Response:
         created_at=time.time(),
         cover_image_path=cover_path,
         cover_image_mime=cover_mime,
+        chapter_intro_delay=chapter_intro_delay,
     )
 
     service.store_pending_job(pending)
@@ -854,6 +860,19 @@ def finalize_job(pending_id: str) -> Response:
     pending = cast(PendingJob, pending)
 
     profiles = serialize_profiles()
+    delay_value = pending.chapter_intro_delay
+    raw_delay = request.form.get("chapter_intro_delay")
+    if raw_delay is not None:
+        raw_normalized = raw_delay.strip()
+        if raw_normalized:
+            try:
+                delay_value = max(0.0, float(raw_normalized))
+            except ValueError:
+                return _render_prepare_page(pending, error="Enter a valid number for the chapter intro delay.")
+        else:
+            delay_value = 0.0
+    pending.chapter_intro_delay = delay_value
+
     overrides: List[Dict[str, Any]] = []
     selected_total = 0
     errors: List[str] = []
@@ -941,6 +960,7 @@ def finalize_job(pending_id: str) -> Response:
         max_subtitle_words=pending.max_subtitle_words,
         cover_image_path=pending.cover_image_path,
         cover_image_mime=pending.cover_image_mime,
+        chapter_intro_delay=pending.chapter_intro_delay,
     )
 
     return redirect(url_for("web.job_detail", job_id=job.id))

abogen/web/service.py

@@ -19,7 +19,7 @@ def _create_set_event() -> threading.Event:
     return event
 
 
-STATE_VERSION = 2
+STATE_VERSION = 3
 
 
 class JobStatus(str, Enum):
@@ -69,6 +69,7 @@ class Job:
     voice_profile: Optional[str] = None
     metadata_tags: Dict[str, str] = field(default_factory=dict)
     max_subtitle_words: int = 50
+    chapter_intro_delay: float = 0.5
     status: JobStatus = JobStatus.PENDING
     started_at: Optional[float] = None
     finished_at: Optional[float] = None
@@ -119,6 +120,7 @@ class Job:
                 "save_as_project": self.save_as_project,
                 "voice_profile": self.voice_profile,
                 "max_subtitle_words": self.max_subtitle_words,
+                "chapter_intro_delay": self.chapter_intro_delay,
             },
             "metadata_tags": dict(self.metadata_tags),
             "chapters": [
@@ -167,6 +169,7 @@ class PendingJob:
     created_at: float
     cover_image_path: Optional[Path] = None
     cover_image_mime: Optional[str] = None
+    chapter_intro_delay: float = 0.5
 
 
 class ConversionService:
@@ -230,6 +233,7 @@ class ConversionService:
         metadata_tags: Optional[Mapping[str, Any]] = None,
         cover_image_path: Optional[Path] = None,
         cover_image_mime: Optional[str] = None,
+        chapter_intro_delay: float = 0.5,
     ) -> Job:
         job_id = uuid.uuid4().hex
         normalized_metadata = self._normalize_metadata_tags(metadata_tags)
@@ -263,6 +267,7 @@ class ConversionService:
             chapters=normalized_chapters,
             cover_image_path=cover_image_path,
             cover_image_mime=cover_image_mime,
+            chapter_intro_delay=chapter_intro_delay,
         )
         with self._lock:
             self._jobs[job_id] = job
@@ -528,6 +533,7 @@ class ConversionService:
             "resume_token": job.resume_token,
             "cover_image_path": str(job.cover_image_path) if job.cover_image_path else None,
             "cover_image_mime": job.cover_image_mime,
+            "chapter_intro_delay": job.chapter_intro_delay,
         }
 
     def _persist_state(self) -> None:
@@ -573,6 +579,7 @@ class ConversionService:
             voice_profile=payload.get("voice_profile"),
             metadata_tags=payload.get("metadata_tags", {}),
             max_subtitle_words=int(payload.get("max_subtitle_words", 50)),
+            chapter_intro_delay=float(payload.get("chapter_intro_delay", 0.5)),
         )
         job.status = JobStatus(payload.get("status", job.status.value))
         job.started_at = payload.get("started_at")

abogen/web/templates/job_detail.html

@@ -23,6 +23,7 @@
         <li><strong>Merge at end:</strong> {{ 'Yes' if job.merge_chapters_at_end else 'No' }}</li>
         <li><strong>Separate chapter format:</strong> {{ job.separate_chapters_format|upper }}</li>
         <li><strong>Silence between chapters:</strong> {{ '%.1f'|format(job.silence_between_chapters) }}s</li>
+        <li><strong>Chapter intro delay:</strong> {{ '%.1f'|format(job.chapter_intro_delay) }}s</li>
         <li><strong>Max words per subtitle:</strong> {{ job.max_subtitle_words }}</li>
         <li><strong>Project folder:</strong> {{ 'Yes' if job.save_as_project else 'No' }}</li>
       </ul>

abogen/web/templates/prepare_job.html

@@ -29,6 +29,10 @@
         <dt>Characters</dt>
         <dd>{{ pending.total_characters|default(0) }}</dd>
       </div>
+      <div>
+        <dt>Chapter intro delay</dt>
+        <dd>{{ '%.1f'|format(pending.chapter_intro_delay) }} seconds</dd>
+      </div>
     </dl>
     {% if pending.metadata_tags %}
     <div class="prepare-metadata">
@@ -105,6 +109,13 @@
       </article>
       {% endfor %}
     </div>
+    <div class="prepare-options">
+      <div class="field">
+        <label for="chapter_intro_delay">Pause after chapter titles (seconds)</label>
+        <input type="number" step="0.1" min="0" id="chapter_intro_delay" name="chapter_intro_delay" value="{{ '%.2f'|format(pending.chapter_intro_delay) }}">
+        <p class="hint">Set to 0 to disable the pause after speaking each chapter title.</p>
+      </div>
+    </div>
     <div class="prepare-actions">
       <button type="submit" class="button">Queue conversion</button>
       <button type="submit" class="button button--ghost" form="cancel-form">Cancel</button>

abogen/web/templates/settings.html

@@ -86,6 +86,11 @@
   <label for="silence_between_chapters">Silence Between Chapters (Seconds)</label>
         <input type="number" step="0.5" min="0" id="silence_between_chapters" name="silence_between_chapters" value="{{ settings.silence_between_chapters }}">
       </div>
+  <div class="field">
+  <label for="chapter_intro_delay">Pause After Chapter Titles (Seconds)</label>
+    <input type="number" step="0.1" min="0" id="chapter_intro_delay" name="chapter_intro_delay" value="{{ '%.2f'|format(settings.chapter_intro_delay) }}">
+    <p class="hint">Inserted between the spoken chapter title and the chapter content. Set to 0 to disable.</p>
+  </div>
       <div class="field">
   <label for="max_subtitle_words">Max Words Per Subtitle Entry</label>
         <input type="number" min="1" max="200" id="max_subtitle_words" name="max_subtitle_words" value="{{ settings.max_subtitle_words }}">