feat: show paragraph progress during Play + warmup first paragraphs at startup

- Play status shows 'Paragraph X / N' instead of blank
- Background warmup pre-generates first paragraph of all 10 stories (stock voice)
- Uses disk cache directly so cancel_pregeneration() won't clear warmup results

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

app.py

@@ -419,6 +419,33 @@ from inference_lfm import get_lfm_model
 get_lfm_model()
 print("[MomsVoice] LFM Q&A model ready. All models preloaded.")
 
+# Pre-generate first paragraph audio for all stories (stock voice) for instant start
+import threading
+
+def _warmup_first_paragraphs():
+    """Background: pre-generate first paragraph of each story with stock voice to disk cache."""
+    import logging as _log
+    from tts import _get_cached_audio, _save_cached_audio, _synthesize_single
+    _logger = _log.getLogger("warmup")
+    for book in mock_books:
+        try:
+            paras = load_paragraphs(book["story_path"])
+            if not paras:
+                continue
+            first_chunks = split_into_chunks(paras[0])
+            for chunk in first_chunks:
+                if _get_cached_audio(chunk, None) is not None:
+                    continue  # Already cached
+                wav, sr = _synthesize_single(chunk, None)
+                _save_cached_audio(chunk, None, wav, sr)
+            _logger.info("Warmed: %s (%d chunks)", book["title"], len(first_chunks))
+        except Exception as e:
+            _logger.warning("Warmup failed for %s: %s", book.get("title", "?"), e)
+    _logger.info("All first paragraphs warmed.")
+
+threading.Thread(target=_warmup_first_paragraphs, daemon=True).start()
+print("[MomsVoice] Background warmup started: pre-generating first paragraph for all stories.")
+
 
 # Gradio Application Core setup
 with gr.Blocks(title="MomsVoice", css=css_code) as demo:
@@ -933,7 +960,7 @@ with gr.Blocks(title="MomsVoice", css=css_code) as demo:
 
         n = len(tts_chunks)
         yield (gr.update(), _status_playing,
-               "",
+               f"<div style='text-align:center;font-size:10px;color:#4ade80;font-family:monospace;margin-top:8px;'>▶ Paragraph 1 / {n_paras}</div>",
                gr.Button(visible=False), gr.Button(visible=True), gr.Button(visible=False),
                gr.Slider(value=0), render_story_text(paras, 0) if paras else "",
                gr.HTML(visible=False),
@@ -950,13 +977,14 @@ with gr.Blocks(title="MomsVoice", css=css_code) as demo:
                 return
             para_idx = min(int(i * n_paras / total), n_paras - 1) if total > 0 else 0
             is_last = (i + 1 >= total)
-            yield ((sample_rate, wav), _status_playing, "",
+            para_status = f"<div style='text-align:center;font-size:10px;color:#4ade80;font-family:monospace;margin-top:8px;'>▶ Paragraph {para_idx + 1} / {n_paras}</div>"
+            yield ((sample_rate, wav), _status_playing, para_status,
                    gr.Button(visible=False), gr.Button(visible=True), gr.Button(visible=True),
                    gr.Slider(value=para_idx), render_story_text(paras, para_idx),
                    gr.HTML(visible=is_last),
                    gr.HTML(visible=False), gr.Group(visible=False), gr.Button(visible=False))
 
-        yield (gr.update(), _status_done, "",
+        yield (gr.update(), _status_done, f"<div style='text-align:center;font-size:10px;color:#94a3b8;font-family:monospace;margin-top:8px;'>✅ Story complete</div>",
                gr.Button(visible=True), gr.Button(visible=False), gr.Button(visible=False),
                gr.Slider(value=max(n_paras - 1, 0)), render_story_text(paras, max(n_paras - 1, 0)),
                gr.HTML(visible=True),