mirror of
https://github.com/denizsafak/abogen.git
synced 2026-07-18 21:50:28 +02:00
feat: Enhance speaker analysis with gender inference and update related tests
This commit is contained in:
JB
@@ -332,7 +332,7 @@ def _apply_prepare_form(
|
||||
"text": chapter.get("text", ""),
|
||||
"enabled": enabled,
|
||||
}
|
||||
entry["characters"] = len(entry["text"])
|
||||
entry["characters"] = calculate_text_length(entry["text"])
|
||||
|
||||
if enabled:
|
||||
if voice_selection.startswith("voice:"):
|
||||
@@ -359,7 +359,7 @@ def _apply_prepare_form(
|
||||
else:
|
||||
entry["voice_formula"] = formula_input
|
||||
entry["resolved_voice"] = formula_input
|
||||
selected_total += len(entry["text"] or "")
|
||||
selected_total += entry["characters"]
|
||||
|
||||
overrides.append(entry)
|
||||
pending.chapters[index] = dict(entry)
|
||||
@@ -1299,7 +1299,7 @@ def enqueue_job() -> ResponseReturnValue:
|
||||
"index": index,
|
||||
"title": chapter.title,
|
||||
"text": chapter.text,
|
||||
"characters": len(chapter.text),
|
||||
"characters": calculate_text_length(chapter.text),
|
||||
"enabled": enabled,
|
||||
}
|
||||
)
|
||||
|
||||
@@ -113,6 +113,7 @@ body {
|
||||
box-shadow: 0 20px 50px rgba(8, 15, 32, 0.35);
|
||||
position: relative;
|
||||
overflow: hidden;
|
||||
z-index: 0;
|
||||
}
|
||||
|
||||
.card::before {
|
||||
@@ -121,6 +122,7 @@ body {
|
||||
inset: 0;
|
||||
background: radial-gradient(circle at 0% 0%, rgba(56, 189, 248, 0.15), transparent 55%);
|
||||
pointer-events: none;
|
||||
z-index: -1;
|
||||
}
|
||||
|
||||
.card__title {
|
||||
@@ -142,11 +144,17 @@ body {
|
||||
align-items: start;
|
||||
}
|
||||
|
||||
.form-grid {
|
||||
grid-template-columns: minmax(0, 420px) minmax(0, 1fr);
|
||||
align-items: start;
|
||||
gap: 2rem;
|
||||
}
|
||||
|
||||
.form-grid > .grid {
|
||||
align-items: start;
|
||||
max-width: 460px;
|
||||
width: min(100%, 460px);
|
||||
justify-self: start;
|
||||
max-width: none;
|
||||
width: 100%;
|
||||
justify-self: stretch;
|
||||
}
|
||||
|
||||
.button {
|
||||
@@ -650,29 +658,50 @@ body {
|
||||
|
||||
.prepare-summary {
|
||||
display: grid;
|
||||
grid-template-columns: repeat(auto-fit, minmax(220px, 1fr));
|
||||
gap: 1.5rem;
|
||||
margin-bottom: 1.5rem;
|
||||
grid-template-columns: minmax(0, 320px) minmax(0, 1fr);
|
||||
gap: 2rem;
|
||||
margin-bottom: 2rem;
|
||||
align-items: start;
|
||||
}
|
||||
|
||||
.prepare-summary dl {
|
||||
.prepare-summary__stats dl {
|
||||
margin: 0;
|
||||
display: grid;
|
||||
gap: 0.6rem;
|
||||
grid-template-columns: repeat(auto-fit, minmax(180px, 1fr));
|
||||
}
|
||||
|
||||
.prepare-summary dt {
|
||||
.prepare-summary__stats dt {
|
||||
font-size: 0.75rem;
|
||||
letter-spacing: 0.04em;
|
||||
text-transform: uppercase;
|
||||
color: var(--muted);
|
||||
}
|
||||
|
||||
.prepare-summary dd {
|
||||
.prepare-summary__stats dd {
|
||||
margin: 0;
|
||||
font-weight: 500;
|
||||
}
|
||||
|
||||
.prepare-summary__insights {
|
||||
display: grid;
|
||||
gap: 1.25rem;
|
||||
align-content: start;
|
||||
}
|
||||
|
||||
.prepare-summary__stats dl > div {
|
||||
display: grid;
|
||||
gap: 0.25rem;
|
||||
}
|
||||
|
||||
.prepare-summary__insights > .prepare-analysis,
|
||||
.prepare-summary__insights > .prepare-metadata {
|
||||
border: 1px solid rgba(148, 163, 184, 0.2);
|
||||
background: rgba(15, 23, 42, 0.35);
|
||||
border-radius: 18px;
|
||||
padding: 1rem 1.25rem;
|
||||
}
|
||||
|
||||
.prepare-speakers {
|
||||
margin-top: 2rem;
|
||||
border-top: 1px solid var(--panel-border);
|
||||
@@ -918,6 +947,26 @@ body {
|
||||
color: var(--accent);
|
||||
}
|
||||
|
||||
@media (max-width: 1024px) {
|
||||
.prepare-summary {
|
||||
grid-template-columns: 1fr;
|
||||
}
|
||||
|
||||
.prepare-summary__stats dl {
|
||||
grid-template-columns: repeat(auto-fit, minmax(160px, 1fr));
|
||||
}
|
||||
}
|
||||
|
||||
@media (max-width: 880px) {
|
||||
.form-grid {
|
||||
grid-template-columns: 1fr;
|
||||
}
|
||||
|
||||
.voice-preview {
|
||||
width: 100%;
|
||||
}
|
||||
}
|
||||
|
||||
.hint {
|
||||
margin: 0;
|
||||
font-size: 0.8rem;
|
||||
@@ -1481,9 +1530,10 @@ progress.progress::-moz-progress-bar {
|
||||
gap: 0.9rem;
|
||||
padding: 1rem 1.25rem;
|
||||
border-radius: 18px;
|
||||
border: 1px solid rgba(148, 163, 184, 0.18);
|
||||
background: rgba(15, 23, 42, 0.35);
|
||||
max-width: 460px;
|
||||
border: 1px solid rgba(148, 163, 184, 0.22);
|
||||
background: rgba(15, 23, 42, 0.6);
|
||||
box-shadow: 0 16px 40px rgba(8, 15, 32, 0.45);
|
||||
width: min(100%, 520px);
|
||||
}
|
||||
|
||||
.field--slider {
|
||||
@@ -1548,6 +1598,13 @@ progress.progress::-moz-progress-bar {
|
||||
border: 1px solid rgba(148, 163, 184, 0.18);
|
||||
}
|
||||
|
||||
.voice-preview textarea {
|
||||
border-radius: 14px;
|
||||
border: 1px solid rgba(148, 163, 184, 0.25);
|
||||
background: rgba(11, 18, 34, 0.85);
|
||||
min-height: 140px;
|
||||
}
|
||||
|
||||
.voice-status {
|
||||
min-height: 1.2rem;
|
||||
font-size: 0.9rem;
|
||||
|
||||
@@ -7,130 +7,138 @@
|
||||
<div class="card__title">Prepare audiobook</div>
|
||||
<p class="card__subtitle">Review the detected chapters, choose which ones to render, and optionally override the voice per chapter before sending the job to the queue.</p>
|
||||
|
||||
{% set analysis = pending.speaker_analysis or {} %}
|
||||
{% set analysis_speakers = analysis.get('speakers', {}) %}
|
||||
{% set show_analysis_prompt = pending.speaker_mode == 'multi' and not pending.analysis_requested %}
|
||||
{% set has_metadata = pending.metadata_tags %}
|
||||
<div class="prepare-summary">
|
||||
<dl>
|
||||
<div>
|
||||
<dt>Source</dt>
|
||||
<dd>{{ pending.original_filename }}</dd>
|
||||
<div class="prepare-summary__stats">
|
||||
<dl>
|
||||
<div>
|
||||
<dt>Source</dt>
|
||||
<dd>{{ pending.original_filename }}</dd>
|
||||
</div>
|
||||
<div>
|
||||
<dt>Language</dt>
|
||||
<dd>{{ options.languages.get(pending.language, pending.language|upper) }}</dd>
|
||||
</div>
|
||||
<div>
|
||||
<dt>Default voice</dt>
|
||||
<dd>{% if pending.voice_profile %}Profile · {{ pending.voice_profile }}{% else %}{{ pending.voice }}{% endif %}</dd>
|
||||
</div>
|
||||
<div>
|
||||
<dt>Total chapters</dt>
|
||||
<dd>{{ pending.chapters|length }}</dd>
|
||||
</div>
|
||||
<div>
|
||||
<dt>Characters</dt>
|
||||
<dd>{{ '{:,}'.format(pending.total_characters|default(0)) }}</dd>
|
||||
</div>
|
||||
<div>
|
||||
<dt>Chapter intro delay</dt>
|
||||
<dd>{{ '%.1f'|format(pending.chapter_intro_delay) }} seconds</dd>
|
||||
</div>
|
||||
<div>
|
||||
<dt>Chunk granularity</dt>
|
||||
<dd>{{ pending.chunk_level|replace('_', ' ')|title }}</dd>
|
||||
</div>
|
||||
<div>
|
||||
<dt>Speaker mode</dt>
|
||||
<dd>{{ pending.speaker_mode|replace('_', ' ')|title }}</dd>
|
||||
</div>
|
||||
<div>
|
||||
<dt>Speaker analysis threshold</dt>
|
||||
<dd>{{ pending.speaker_analysis_threshold }} {{ 'mention' if pending.speaker_analysis_threshold == 1 else 'mentions' }}</dd>
|
||||
</div>
|
||||
<div>
|
||||
<dt>EPUB 3 package</dt>
|
||||
<dd>{% if pending.generate_epub3 %}Enabled{% else %}Disabled{% endif %}</dd>
|
||||
</div>
|
||||
</dl>
|
||||
</div>
|
||||
{% if analysis_speakers or show_analysis_prompt or has_metadata %}
|
||||
<div class="prepare-summary__insights">
|
||||
{% if analysis_speakers %}
|
||||
{% set active = namespace(items=[]) %}
|
||||
{% for sid, payload in analysis_speakers.items() %}
|
||||
{% if sid != 'narrator' and not payload.get('suppressed') %}
|
||||
{% set _ = active.items.append(payload) %}
|
||||
{% endif %}
|
||||
{% endfor %}
|
||||
<div class="prepare-analysis">
|
||||
<h2>Detected speakers</h2>
|
||||
{% if active.items %}
|
||||
<ul>
|
||||
{% for speaker in active.items|sort(attribute='label') %}
|
||||
<li><strong>{{ speaker.label }}</strong> · {{ speaker.count }} lines · confidence {{ speaker.confidence|title }}</li>
|
||||
{% endfor %}
|
||||
</ul>
|
||||
{% else %}
|
||||
<p>No additional speakers met the threshold yet. All dialogue will use the narrator voice.</p>
|
||||
{% endif %}
|
||||
</div>
|
||||
<div>
|
||||
<dt>Language</dt>
|
||||
<dd>{{ options.languages.get(pending.language, pending.language|upper) }}</dd>
|
||||
{% elif show_analysis_prompt %}
|
||||
<div class="prepare-analysis">
|
||||
<h2>Detected speakers</h2>
|
||||
<p>Press <strong>Analyze speakers</strong> after selecting chapters to discover recurring voices.</p>
|
||||
</div>
|
||||
<div>
|
||||
<dt>Default voice</dt>
|
||||
<dd>{% if pending.voice_profile %}Profile · {{ pending.voice_profile }}{% else %}{{ pending.voice }}{% endif %}</dd>
|
||||
</div>
|
||||
<div>
|
||||
<dt>Total chapters</dt>
|
||||
<dd>{{ pending.chapters|length }}</dd>
|
||||
</div>
|
||||
<div>
|
||||
<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>
|
||||
<div>
|
||||
<dt>Chunk granularity</dt>
|
||||
<dd>{{ pending.chunk_level|replace('_', ' ')|title }}</dd>
|
||||
</div>
|
||||
<div>
|
||||
<dt>Speaker mode</dt>
|
||||
<dd>{{ pending.speaker_mode|replace('_', ' ')|title }}</dd>
|
||||
</div>
|
||||
<div>
|
||||
<dt>Speaker analysis threshold</dt>
|
||||
<dd>{{ pending.speaker_analysis_threshold }} {{ 'mention' if pending.speaker_analysis_threshold == 1 else 'mentions' }}</dd>
|
||||
</div>
|
||||
<div>
|
||||
<dt>EPUB 3 package</dt>
|
||||
<dd>{% if pending.generate_epub3 %}Enabled{% else %}Disabled{% endif %}</dd>
|
||||
</div>
|
||||
</dl>
|
||||
{% set analysis = pending.speaker_analysis or {} %}
|
||||
{% set analysis_speakers = analysis.get('speakers', {}) %}
|
||||
{% if analysis_speakers %}
|
||||
{% set active = namespace(items=[]) %}
|
||||
{% for sid, payload in analysis_speakers.items() %}
|
||||
{% if sid != 'narrator' and not payload.get('suppressed') %}
|
||||
{% set _ = active.items.append(payload) %}
|
||||
{% endif %}
|
||||
{% endfor %}
|
||||
<div class="prepare-analysis">
|
||||
<h2>Detected speakers</h2>
|
||||
{% if active.items %}
|
||||
<ul>
|
||||
{% for speaker in active.items|sort(attribute='label') %}
|
||||
<li><strong>{{ speaker.label }}</strong> · {{ speaker.count }} lines · confidence {{ speaker.confidence|title }}</li>
|
||||
{% endfor %}
|
||||
</ul>
|
||||
{% else %}
|
||||
<p>No additional speakers met the threshold yet. All dialogue will use the narrator voice.</p>
|
||||
{% if has_metadata %}
|
||||
<div class="prepare-metadata">
|
||||
<h2>Metadata</h2>
|
||||
<ul>
|
||||
{% for key, value in pending.metadata_tags.items() %}
|
||||
<li><strong>{{ key|replace('_', ' ')|title }}:</strong> {{ value }}</li>
|
||||
{% endfor %}
|
||||
</ul>
|
||||
</div>
|
||||
{% endif %}
|
||||
</div>
|
||||
{% elif pending.speaker_mode == 'multi' and not pending.analysis_requested %}
|
||||
<div class="prepare-analysis">
|
||||
<h2>Detected speakers</h2>
|
||||
<p>Press <strong>Analyze speakers</strong> after selecting chapters to discover recurring voices.</p>
|
||||
</div>
|
||||
{% endif %}
|
||||
{% set roster = pending.speakers or {} %}
|
||||
{% if roster %}
|
||||
{% set preview_template = options.speaker_pronunciation_sentence or "This is {{name}} speaking." %}
|
||||
<div class="prepare-speakers">
|
||||
<h2>Speaker pronunciation guide</h2>
|
||||
<p class="hint">Add a phonetic spelling (IPA or plain text) so pronunciations sound right. Leave blank to use the written label.</p>
|
||||
<ul class="speaker-list">
|
||||
{% for speaker_id, speaker in roster.items() %}
|
||||
{% set spoken_name = speaker.pronunciation or speaker.label %}
|
||||
{% set preview_text = preview_template | replace("{{name}}", spoken_name) %}
|
||||
<li class="speaker-list__item">
|
||||
<div class="speaker-list__header">
|
||||
<span class="speaker-list__name">{{ speaker.label }}</span>
|
||||
<button type="button"
|
||||
class="icon-button speaker-list__preview"
|
||||
data-role="speaker-preview"
|
||||
data-speaker-id="{{ speaker_id }}"
|
||||
data-preview-text="{{ preview_text|e }}"
|
||||
data-language="{{ pending.language }}"
|
||||
data-voice="{{ speaker.resolved_voice or speaker.voice_formula or speaker.voice or pending.voice }}"
|
||||
data-speed="{{ '%.2f'|format(pending.speed) }}"
|
||||
data-use-gpu="{{ 'true' if pending.use_gpu else 'false' }}"
|
||||
aria-label="Preview pronunciation for {{ speaker.label }}"
|
||||
title="Preview pronunciation">
|
||||
🔊
|
||||
</button>
|
||||
</div>
|
||||
<label class="speaker-list__field" for="speaker-{{ speaker_id }}-pronunciation">
|
||||
<span>Pronunciation</span>
|
||||
<input type="text"
|
||||
id="speaker-{{ speaker_id }}-pronunciation"
|
||||
name="speaker-{{ speaker_id }}-pronunciation"
|
||||
value="{{ speaker.pronunciation or '' }}"
|
||||
placeholder="{{ speaker.label }}">
|
||||
</label>
|
||||
{% if speaker.get('analysis_count') %}
|
||||
<p class="hint speaker-list__stats">{{ speaker.analysis_count }} detected lines · confidence {{ speaker.analysis_confidence|default('low')|title }}</p>
|
||||
{% endif %}
|
||||
</li>
|
||||
{% endfor %}
|
||||
</ul>
|
||||
</div>
|
||||
{% endif %}
|
||||
{% if pending.metadata_tags %}
|
||||
<div class="prepare-metadata">
|
||||
<h2>Metadata</h2>
|
||||
<ul>
|
||||
{% for key, value in pending.metadata_tags.items() %}
|
||||
<li><strong>{{ key|replace('_', ' ')|title }}:</strong> {{ value }}</li>
|
||||
{% endfor %}
|
||||
</ul>
|
||||
</div>
|
||||
{% endif %}
|
||||
</div>
|
||||
{% set roster = pending.speakers or {} %}
|
||||
{% if roster %}
|
||||
{% set preview_template = options.speaker_pronunciation_sentence or "This is {{name}} speaking." %}
|
||||
<div class="prepare-speakers">
|
||||
<h2>Speaker pronunciation guide</h2>
|
||||
<p class="hint">Add a phonetic spelling (IPA or plain text) so pronunciations sound right. Leave blank to use the written label.</p>
|
||||
<ul class="speaker-list">
|
||||
{% for speaker_id, speaker in roster.items() %}
|
||||
{% set spoken_name = speaker.pronunciation or speaker.label %}
|
||||
{% set preview_text = preview_template | replace("{{name}}", spoken_name) %}
|
||||
<li class="speaker-list__item">
|
||||
<div class="speaker-list__header">
|
||||
<span class="speaker-list__name">{{ speaker.label }}</span>
|
||||
<button type="button"
|
||||
class="icon-button speaker-list__preview"
|
||||
data-role="speaker-preview"
|
||||
data-speaker-id="{{ speaker_id }}"
|
||||
data-preview-text="{{ preview_text|e }}"
|
||||
data-language="{{ pending.language }}"
|
||||
data-voice="{{ speaker.resolved_voice or speaker.voice_formula or speaker.voice or pending.voice }}"
|
||||
data-speed="{{ '%.2f'|format(pending.speed) }}"
|
||||
data-use-gpu="{{ 'true' if pending.use_gpu else 'false' }}"
|
||||
aria-label="Preview pronunciation for {{ speaker.label }}"
|
||||
title="Preview pronunciation">
|
||||
🔊
|
||||
</button>
|
||||
</div>
|
||||
<label class="speaker-list__field" for="speaker-{{ speaker_id }}-pronunciation">
|
||||
<span>Pronunciation</span>
|
||||
<input type="text"
|
||||
id="speaker-{{ speaker_id }}-pronunciation"
|
||||
name="speaker-{{ speaker_id }}-pronunciation"
|
||||
value="{{ speaker.pronunciation or '' }}"
|
||||
placeholder="{{ speaker.label }}">
|
||||
</label>
|
||||
{% if speaker.get('analysis_count') %}
|
||||
<p class="hint speaker-list__stats">{{ speaker.analysis_count }} detected lines · confidence {{ speaker.analysis_confidence|default('low')|title }}</p>
|
||||
{% endif %}
|
||||
</li>
|
||||
{% endfor %}
|
||||
</ul>
|
||||
</div>
|
||||
{% endif %}
|
||||
|
||||
{% if error %}
|
||||
<div class="alert alert--error">{{ error }}</div>
|
||||
@@ -160,7 +168,7 @@
|
||||
</label>
|
||||
</div>
|
||||
<div class="chapter-card__metrics">
|
||||
{{ chapter.characters }} characters
|
||||
{{ '{:,}'.format(chapter.characters) }} characters
|
||||
</div>
|
||||
</header>
|
||||
<div class="chapter-card__body">
|
||||
|
||||
Reference in New Issue
Block a user