fix(vtm): install STT deps in Docker image, enable whisper

voice-to-mermaid/README.md

@@ -78,7 +78,7 @@ It is volume-mounted in Docker, so changes take effect on restart without rebuil
 | `ollama.default_model` | `qwen3:8b` | Ollama model for diagram generation |
 | `ollama.model_filter` | _(list)_ | Model name prefixes shown in the UI picker |
 | `openai.model` | `gpt-4o-mini` | OpenAI-compatible model name |
-| `whisper.enabled` | `false` | Set `true` after installing `requirements.stt.txt`; graceful no-op if import fails |
+| `whisper.enabled` | `true` | STT enabled by default (deps installed in Docker image); graceful no-op if import fails |
 | `whisper.model` | `medium.en` | Whisper model size |
 | `whisper.device` | `auto` | `auto` / `cpu` / `cuda` |
 | `paths.prompt` | `prompts/mermaid.txt` | Path to LLM prompt template |

voice-to-mermaid/backend/Dockerfile

@@ -4,8 +4,8 @@ FROM python:3.12-slim
 RUN apt-get update && apt-get install -y ffmpeg && rm -rf /var/lib/apt/lists/*
 
 WORKDIR /app
-COPY requirements.txt .
-RUN pip install --no-cache-dir -r requirements.txt
+COPY requirements.txt requirements.stt.txt ./
+RUN pip install --no-cache-dir -r requirements.stt.txt
 
 COPY . .
 

voice-to-mermaid/backend/config.yaml

@@ -18,9 +18,9 @@ openai:
   model: "gpt-4o-mini"
 
 whisper:
-  # Set true only after installing requirements.stt.txt (or adding it to Dockerfile).
+  # Enabled: requirements.stt.txt is now installed in the Dockerfile.
   # If import fails, _stt_available stays False and /v1/config returns stt_enabled: false — no crash.
-  enabled: false
+  enabled: true
   model: "medium.en"
   language: "en"
   device: "auto"   # "auto", "cuda", "cpu" — maps to WhisperLiveKitConfig backend

voice-to-mermaid/backend/main.py

@@ -333,7 +333,7 @@ async def health():
 @app.get("/v1/config")
 async def config():
     global _ollama_models_cache
-    if _ollama_models_cache is None:
+    if not _ollama_models_cache:
         _ollama_models_cache = await _fetch_ollama_models()
     return {
         "ollama_url": OLLAMA_URL,
@@ -357,9 +357,6 @@ async def refresh_models():
 
 @app.websocket("/ws/mermaid")
 async def ws_mermaid(websocket: WebSocket):
-    if _stt_available:
-        from whisperlivekit import AudioProcessor
-
     await websocket.accept()
 
     llm_mode: str = "ollama"
@@ -456,6 +453,7 @@ async def _consume_results(results_generator, line_count_ref: list[int]) -> None
     line_count_ref = [0]
 
     if _stt_available:
+        from whisperlivekit import AudioProcessor
         audio_processor = AudioProcessor(transcription_engine=_transcription_engine)
         results_generator = await audio_processor.create_tasks()
         results_task = asyncio.create_task(_consume_results(results_generator, line_count_ref))