feat: Expand debug TTS samples with additional cases and implement validation test for minimum sample counts

abogen/debug_tts_samples.py

@@ -21,54 +21,254 @@ class DebugTTSSample:
 DEBUG_TTS_SAMPLES: Sequence[DebugTTSSample] = (
     DebugTTSSample(
         code="APOS_001",
-        label="Apostrophes & contractions",
+        label="Apostrophes & contractions (1)",
         text="It's a beautiful day, isn't it? Let's see what we'll do.",
     ),
+    DebugTTSSample(
+        code="APOS_002",
+        label="Apostrophes & contractions (2)",
+        text="I'm sure you're ready; we'd better go before it's too late.",
+    ),
+    DebugTTSSample(
+        code="APOS_003",
+        label="Apostrophes & contractions (3)",
+        text="He'll say it's fine, but I can't promise it'll work.",
+    ),
+    DebugTTSSample(
+        code="APOS_004",
+        label="Apostrophes & contractions (4)",
+        text="They've done it, and I'd agree they've earned it.",
+    ),
+    DebugTTSSample(
+        code="APOS_005",
+        label="Apostrophes & contractions (5)",
+        text="She's here, we're late, they're waiting, and you're right.",
+    ),
     DebugTTSSample(
         code="POS_001",
-        label="Plural possessives",
+        label="Plural possessives (1)",
         text="The dogs' bowls were empty, but the boss's office was quiet.",
     ),
+    DebugTTSSample(
+        code="POS_002",
+        label="Plural possessives (2)",
+        text="The teachers' lounge was closed during the students' exams.",
+    ),
+    DebugTTSSample(
+        code="POS_003",
+        label="Plural possessives (3)",
+        text="The actresses' roles changed, and the directors' notes piled up.",
+    ),
+    DebugTTSSample(
+        code="POS_004",
+        label="Plural possessives (4)",
+        text="The Joneses' car was parked by the neighbors' fence.",
+    ),
+    DebugTTSSample(
+        code="POS_005",
+        label="Plural possessives (5)",
+        text="The bosses' meeting ended before the witnesses' statements began.",
+    ),
     DebugTTSSample(
         code="NUM_001",
-        label="Grouped numbers",
+        label="Grouped numbers (1)",
         text="There are 1,234 apples, 56 oranges, and 7.89 liters of juice.",
     ),
+    DebugTTSSample(
+        code="NUM_002",
+        label="Grouped numbers (2)",
+        text="The population is 10,000,000 and the area is 123.45 square miles.",
+    ),
+    DebugTTSSample(
+        code="NUM_003",
+        label="Grouped numbers (3)",
+        text="Set the timer for 0.5 seconds, then wait 2.0 minutes.",
+    ),
+    DebugTTSSample(
+        code="NUM_004",
+        label="Grouped numbers (4)",
+        text="We measured 3.1415 radians and wrote down 2,718.28 as well.",
+    ),
+    DebugTTSSample(
+        code="NUM_005",
+        label="Grouped numbers (5)",
+        text="The sequence is 1, 2, 3, 4, 5, and then 13.",
+    ),
     DebugTTSSample(
         code="YEAR_001",
-        label="Years and decades",
+        label="Years and decades (1)",
         text="In 1999, people said the '90s were over.",
     ),
+    DebugTTSSample(
+        code="YEAR_002",
+        label="Years and decades (2)",
+        text="In 2001, the show premiered; by 2010 it was everywhere.",
+    ),
+    DebugTTSSample(
+        code="YEAR_003",
+        label="Years and decades (3)",
+        text="The 1980s were loud, and the 1970s were groovy.",
+    ),
+    DebugTTSSample(
+        code="YEAR_004",
+        label="Years and decades (4)",
+        text="She loved the '80s, but he preferred the '60s.",
+    ),
+    DebugTTSSample(
+        code="YEAR_005",
+        label="Years and decades (5)",
+        text="In 2024, we looked back at 2020 and planned for 2030.",
+    ),
     DebugTTSSample(
         code="DATE_001",
-        label="ISO dates",
+        label="Dates (1)",
         text="On 2023-01-01, we celebrated the new year.",
     ),
+    DebugTTSSample(
+        code="DATE_002",
+        label="Dates (2)",
+        text="The deadline is 1999-12-31 at midnight.",
+    ),
+    DebugTTSSample(
+        code="DATE_003",
+        label="Dates (3)",
+        text="Leap day happens on 2024-02-29.",
+    ),
+    DebugTTSSample(
+        code="DATE_004",
+        label="Dates (4)",
+        text="Some formats look like 01/02/2003 and can be ambiguous.",
+    ),
+    DebugTTSSample(
+        code="DATE_005",
+        label="Dates (5)",
+        text="We met on March 5, 2020 and again on Apr. 7, 2021.",
+    ),
     DebugTTSSample(
         code="CUR_001",
-        label="Currency symbols",
+        label="Currency symbols (1)",
         text="The price is $10.50, but it was £8.00 yesterday.",
     ),
+    DebugTTSSample(
+        code="CUR_002",
+        label="Currency symbols (2)",
+        text="Tickets cost €12, and the fine was $0.99.",
+    ),
+    DebugTTSSample(
+        code="CUR_003",
+        label="Currency symbols (3)",
+        text="The bill was ¥500 and the refund was $-3.25.",
+    ),
+    DebugTTSSample(
+        code="CUR_004",
+        label="Currency symbols (4)",
+        text="He paid £1,234.56 for the instrument.",
+    ),
+    DebugTTSSample(
+        code="CUR_005",
+        label="Currency symbols (5)",
+        text="The subscription is $5 per month, or $50 per year.",
+    ),
     DebugTTSSample(
         code="TITLE_001",
-        label="Titles and abbreviations",
+        label="Titles and abbreviations (1)",
         text="Dr. Smith lives on Elm St. near the U.S. border.",
     ),
+    DebugTTSSample(
+        code="TITLE_002",
+        label="Titles and abbreviations (2)",
+        text="Mr. and Mrs. Doe met Prof. Adams at 5 p.m.",
+    ),
+    DebugTTSSample(
+        code="TITLE_003",
+        label="Titles and abbreviations (3)",
+        text="Gen. Lee spoke to Sgt. Rivera on Main St.",
+    ),
+    DebugTTSSample(
+        code="TITLE_004",
+        label="Titles and abbreviations (4)",
+        text="The report came from the U.K. office, not the U.S.A. team.",
+    ),
+    DebugTTSSample(
+        code="TITLE_005",
+        label="Titles and abbreviations (5)",
+        text="St. John's is different from St. Louis.",
+    ),
     DebugTTSSample(
         code="PUNC_001",
-        label="Terminal punctuation",
+        label="Terminal punctuation (1)",
         text="This sentence ends without punctuation",
     ),
+    DebugTTSSample(
+        code="PUNC_002",
+        label="Terminal punctuation (2)",
+        text="An ellipsis is already present...",
+    ),
+    DebugTTSSample(
+        code="PUNC_003",
+        label="Terminal punctuation (3)",
+        text="A question without a mark",
+    ),
+    DebugTTSSample(
+        code="PUNC_004",
+        label="Terminal punctuation (4)",
+        text="An exclamation without a bang",
+    ),
+    DebugTTSSample(
+        code="PUNC_005",
+        label="Terminal punctuation (5)",
+        text='A quote ends here"',
+    ),
     DebugTTSSample(
         code="QUOTE_001",
-        label="ALL CAPS inside quotes",
+        label="ALL CAPS inside quotes (1)",
         text='He shouted, "THIS IS IMPORTANT!" and then whispered, "ok."',
     ),
+    DebugTTSSample(
+        code="QUOTE_002",
+        label="ALL CAPS inside quotes (2)",
+        text='She said, "NO WAY", but he replied, "maybe".',
+    ),
+    DebugTTSSample(
+        code="QUOTE_003",
+        label="ALL CAPS inside quotes (3)",
+        text='The sign read "DO NOT ENTER" and the note read "pls knock".',
+    ),
+    DebugTTSSample(
+        code="QUOTE_004",
+        label="ALL CAPS inside quotes (4)",
+        text='He muttered, "OK", then yelled, "STOP!"',
+    ),
+    DebugTTSSample(
+        code="QUOTE_005",
+        label="ALL CAPS inside quotes (5)",
+        text='They chanted, "USA!" and someone wrote "idk".',
+    ),
     DebugTTSSample(
         code="FOOT_001",
-        label="Footnote indicators",
+        label="Footnote indicators (1)",
         text="This is a sentence with a footnote[1] and another[12].",
     ),
+    DebugTTSSample(
+        code="FOOT_002",
+        label="Footnote indicators (2)",
+        text="Some books use multiple footnotes like this[2][3] in a row.",
+    ),
+    DebugTTSSample(
+        code="FOOT_003",
+        label="Footnote indicators (3)",
+        text="A footnote can appear mid-sentence[4] and continue afterward.",
+    ),
+    DebugTTSSample(
+        code="FOOT_004",
+        label="Footnote indicators (4)",
+        text="Edge cases include [0] or very large indices like [1234].",
+    ),
+    DebugTTSSample(
+        code="FOOT_005",
+        label="Footnote indicators (5)",
+        text="Sometimes a footnote follows punctuation.[5] Sometimes it doesn't[6]",
+    ),
 )
 
 

abogen/web/templates/settings.html

@@ -11,6 +11,14 @@
   <div class="alert alert--success">Settings saved successfully.</div>
   {% endif %}
 
+  {% with messages = get_flashed_messages(with_categories=True) %}
+  {% if messages %}
+    {% for category, message in messages %}
+      <div class="alert {% if category == 'error' %}alert--error{% else %}alert--success{% endif %}">{{ message }}</div>
+    {% endfor %}
+  {% endif %}
+  {% endwith %}
+
   <div class="settings-layout">
     <nav class="settings-nav" aria-label="Settings sections">
       <button type="button" class="settings-nav__item is-active" data-section="narration">Narration</button>
@@ -480,8 +488,8 @@
             <p class="hint">Generate a set of WAV files from a purpose-built EPUB containing code-tagged examples. When something sounds wrong, report the code (e.g. <code>NUM_001</code>) to pinpoint the failing transformation.</p>
 
             <div class="field field--stack">
-              <button type="submit" class="button" formaction="{{ url_for('settings.run_debug_wavs') }}" formmethod="post">Generate debug WAVs</button>
+              <button type="submit" class="button" form="debug-tts-form">Generate debug WAVs</button>
-              <p class="hint">Uses your current Settings defaults (voice, language, speed, GPU).</p>
+              <p class="hint">Uses your current Settings defaults (voice, language, speed, GPU). If generation fails, an error will appear at the top of this page.</p>
             </div>
 
             {% if debug_manifest and debug_manifest.artifacts %}
@@ -582,6 +590,8 @@
         <button type="submit" class="button">Save Settings</button>
       </div>
     </form>
+
+    <form id="debug-tts-form" action="{{ url_for('settings.run_debug_wavs') }}" method="post"></form>
   </div>
 </section>
 {% endblock %}

tests/test_debug_tts_samples.py

@@ -95,3 +95,27 @@ def test_settings_debug_route_writes_manifest(tmp_path, monkeypatch):
         filenames = {item["filename"] for item in manifest.get("artifacts", [])}
         assert "overall.wav" in filenames
         assert any(name.startswith("case_") and name.endswith(".wav") for name in filenames)
+
+
+def test_debug_samples_have_minimum_per_category():
+    prefixes = {
+        "APOS": 5,
+        "POS": 5,
+        "NUM": 5,
+        "YEAR": 5,
+        "DATE": 5,
+        "CUR": 5,
+        "TITLE": 5,
+        "PUNC": 5,
+        "QUOTE": 5,
+        "FOOT": 5,
+    }
+
+    counts = {prefix: 0 for prefix in prefixes}
+    for sample in DEBUG_TTS_SAMPLES:
+        prefix = sample.code.split("_", 1)[0]
+        if prefix in counts:
+            counts[prefix] += 1
+
+    for prefix, minimum in prefixes.items():
+        assert counts[prefix] >= minimum