remove mistral deadcode; notebooks untouched

.env.template

@@ -90,16 +90,12 @@ CHAT_TEMPERATURE=0.7
 # SPEECH-TO-TEXT CONFIGURATION
 # ========================================
 
-# Primary transcription provider: deepgram, mistral, or parakeet
+# Primary transcription provider: deepgram or parakeet
 TRANSCRIPTION_PROVIDER=deepgram
 
 # Deepgram configuration
 DEEPGRAM_API_KEY=your-deepgram-key-here
 
-# Mistral configuration (when TRANSCRIPTION_PROVIDER=mistral)
-MISTRAL_API_KEY=your-mistral-key-here
-MISTRAL_MODEL=voxtral-mini-2507
-
 # Parakeet ASR configuration (when TRANSCRIPTION_PROVIDER=parakeet)
 PARAKEET_ASR_URL=http://host.docker.internal:8767
 

CLAUDE.md

@@ -18,7 +18,7 @@ This supports a comprehensive web dashboard for management.
 Chronicle includes an **interactive setup wizard** for easy configuration. The wizard guides you through:
 - Service selection (backend + optional services)
 - Authentication setup (admin account, JWT secrets)
-- Transcription provider configuration (Deepgram, Mistral, or offline ASR)
+- Transcription provider configuration (Deepgram or offline ASR)
 - LLM provider setup (OpenAI or Ollama)
 - Memory provider selection (Chronicle Native with Qdrant or OpenMemory MCP)
 - Network configuration and HTTPS setup
@@ -184,12 +184,12 @@ docker compose up --build
 ## Architecture Overview
 
 ### Key Components
-- **Audio Pipeline**: Real-time Opus/PCM → Application-level processing → Deepgram/Mistral transcription → memory extraction
+- **Audio Pipeline**: Real-time Opus/PCM → Application-level processing → Deepgram transcription → memory extraction
 - **Wyoming Protocol**: WebSocket communication uses Wyoming protocol (JSONL + binary) for structured audio sessions
 - **Unified Pipeline**: Job-based tracking system for all audio processing (WebSocket and file uploads)
 - **Job Tracker**: Tracks pipeline jobs with stage events (audio → transcription → memory) and completion status
 - **Task Management**: BackgroundTaskManager tracks all async tasks to prevent orphaned processes
-- **Unified Transcription**: Deepgram/Mistral transcription with fallback to offline ASR services
+- **Unified Transcription**: Deepgram transcription with fallback to offline ASR services
 - **Memory System**: Pluggable providers (Chronicle native or OpenMemory MCP)
 - **Authentication**: Email-based login with MongoDB ObjectId user system
 - **Client Management**: Auto-generated client IDs as `{user_id_suffix}-{device_name}`, centralized ClientManager
@@ -205,7 +205,7 @@ Required:
 
 Recommended:
   - Vector Storage: Qdrant (Chronicle provider) or OpenMemory MCP server
-  - Transcription: Deepgram, Mistral, or offline ASR services
+  - Transcription: Deepgram or offline ASR services
 
 Optional:
   - Parakeet ASR: Offline transcription service
@@ -329,12 +329,7 @@ Chronicle supports multiple transcription services:
 TRANSCRIPTION_PROVIDER=deepgram
 DEEPGRAM_API_KEY=your-deepgram-key-here
 
-# Option 2: Mistral (Voxtral models)
-TRANSCRIPTION_PROVIDER=mistral
-MISTRAL_API_KEY=your-mistral-key-here
-MISTRAL_MODEL=voxtral-mini-2507
-
-# Option 3: Local ASR (Parakeet)
+# Option 2: Local ASR (Parakeet)
 PARAKEET_ASR_URL=http://host.docker.internal:8767
 ```
 

Docs/getting-started.md

@@ -36,7 +36,7 @@ cd backends/advanced
 
 **The setup wizard will guide you through:**
 - **Authentication**: Admin email/password setup
-- **Transcription Provider**: Choose Deepgram, Mistral, or Offline (Parakeet)
+- **Transcription Provider**: Choose Deepgram or Offline (Parakeet)
 - **LLM Provider**: Choose OpenAI or Ollama for memory extraction
 - **Memory Provider**: Choose Chronicle Native or OpenMemory MCP
 - **Optional Services**: Speaker Recognition and other extras
@@ -52,14 +52,13 @@ cd backends/advanced
 Admin email [admin@example.com]: john@company.com
 Admin password (min 8 chars): ********
 
-► Speech-to-Text Configuration  
+► Speech-to-Text Configuration
 -------------------------------
 Choose your transcription provider:
   1) Deepgram (recommended - high quality, requires API key)
-  2) Mistral (Voxtral models - requires API key) 
-  3) Offline (Parakeet ASR - requires GPU, runs locally)
-  4) None (skip transcription setup)
-Enter choice (1-4) [1]: 1
+  2) Offline (Parakeet ASR - requires GPU, runs locally)
+  3) None (skip transcription setup)
+Enter choice (1-3) [1]: 1
 
 Get your API key from: https://console.deepgram.com/
 Deepgram API key: dg_xxxxxxxxxxxxx
@@ -154,20 +153,14 @@ OLLAMA_BASE_URL=http://ollama:11434
 TRANSCRIPTION_PROVIDER=deepgram
 DEEPGRAM_API_KEY=your-deepgram-api-key-here
 
-# Option 2: Mistral (Voxtral models for transcription)
-TRANSCRIPTION_PROVIDER=mistral
-MISTRAL_API_KEY=your-mistral-api-key-here
-MISTRAL_MODEL=voxtral-mini-2507
-
-# Option 3: Local ASR service  
+# Option 2: Local ASR service
 PARAKEET_ASR_URL=http://host.docker.internal:8080
 ```
 
 **Important Notes:**
 - **OpenAI is strongly recommended** for LLM processing as it provides much better memory extraction and eliminates JSON parsing errors
 - **TRANSCRIPTION_PROVIDER** determines which service to use:
   - `deepgram`: Uses Deepgram's Nova-3 model for high-quality transcription
-  - `mistral`: Uses Mistral's Voxtral models for transcription
   - If not set, system falls back to offline ASR service
 - The system requires either online API keys or offline ASR service configuration
 
@@ -312,7 +305,6 @@ curl -X POST "http://localhost:8000/api/process-audio-files" \
 
 ### Transcription Options
 - **Deepgram API**: Cloud-based batch processing, high accuracy (recommended)
-- **Mistral API**: Voxtral models for transcription with REST API processing
 - **Self-hosted ASR**: Local Wyoming protocol services with real-time processing
 - **Collection timeout**: 1.5 minute collection for optimal online processing quality
 
@@ -407,7 +399,6 @@ uv sync --group (whatever group you want to sync)
 
 **Transcription Issues:**
 - **Deepgram**: Verify API key is valid and `TRANSCRIPTION_PROVIDER=deepgram`
-- **Mistral**: Verify API key is valid and `TRANSCRIPTION_PROVIDER=mistral`
 - **Self-hosted**: Ensure ASR service is running on port 8765
 - Check transcription service connection in health endpoint
 

backends/advanced/Docs/memory-configuration-guide.md

@@ -65,7 +65,7 @@ memory:
 - **Embeddings**: `text-embedding-3-small`, `text-embedding-3-large`
 
 #### Ollama Models (Local)
-- **LLM**: `llama3`, `mistral`, `qwen2.5`
+- **LLM**: `llama3`, `qwen2.5`
 - **Embeddings**: `nomic-embed-text`, `all-minilm`
 
 ## Hot Reload

backends/advanced/Docs/quickstart.md

@@ -34,7 +34,7 @@ cd backends/advanced
 
 **The setup wizard will guide you through:**
 - **Authentication**: Admin email/password setup
-- **Transcription Provider**: Choose Deepgram, Mistral, or Offline (Parakeet)
+- **Transcription Provider**: Choose Deepgram or Offline (Parakeet)
 - **LLM Provider**: Choose OpenAI or Ollama for memory extraction
 - **Memory Provider**: Choose Chronicle Native or OpenMemory MCP
 - **Optional Services**: Speaker Recognition and other extras
@@ -50,14 +50,13 @@ cd backends/advanced
 Admin email [admin@example.com]: john@company.com
 Admin password (min 8 chars): ********
 
-► Speech-to-Text Configuration  
+► Speech-to-Text Configuration
 -------------------------------
 Choose your transcription provider:
   1) Deepgram (recommended - high quality, requires API key)
-  2) Mistral (Voxtral models - requires API key) 
-  3) Offline (Parakeet ASR - requires GPU, runs locally)
-  4) None (skip transcription setup)
-Enter choice (1-4) [1]: 1
+  2) Offline (Parakeet ASR - requires GPU, runs locally)
+  3) None (skip transcription setup)
+Enter choice (1-3) [1]: 1
 
 Get your API key from: https://console.deepgram.com/
 Deepgram API key: dg_xxxxxxxxxxxxx
@@ -152,20 +151,14 @@ OLLAMA_BASE_URL=http://ollama:11434
 TRANSCRIPTION_PROVIDER=deepgram
 DEEPGRAM_API_KEY=your-deepgram-api-key-here
 
-# Option 2: Mistral (Voxtral models for transcription)
-TRANSCRIPTION_PROVIDER=mistral
-MISTRAL_API_KEY=your-mistral-api-key-here
-MISTRAL_MODEL=voxtral-mini-2507
-
-# Option 3: Local ASR service  
+# Option 2: Local ASR service
 PARAKEET_ASR_URL=http://host.docker.internal:8080
 ```
 
 **Important Notes:**
 - **OpenAI is strongly recommended** for LLM processing as it provides much better memory extraction and eliminates JSON parsing errors
 - **TRANSCRIPTION_PROVIDER** determines which service to use:
   - `deepgram`: Uses Deepgram's Nova-3 model for high-quality transcription
-  - `mistral`: Uses Mistral's Voxtral models for transcription
   - If not set, system falls back to offline ASR service
 - The system requires either online API keys or offline ASR service configuration
 
@@ -310,7 +303,6 @@ curl -X POST "http://localhost:8000/api/audio/upload" \
 
 ### Transcription Options
 - **Deepgram API**: Cloud-based batch processing, high accuracy (recommended)
-- **Mistral API**: Voxtral models for transcription with REST API processing
 - **Self-hosted ASR**: Local Wyoming protocol services with real-time processing
 - **Collection timeout**: 1.5 minute collection for optimal online processing quality
 
@@ -405,7 +397,6 @@ uv sync --group (whatever group you want to sync)
 
 **Transcription Issues:**
 - **Deepgram**: Verify API key is valid and `TRANSCRIPTION_PROVIDER=deepgram`
-- **Mistral**: Verify API key is valid and `TRANSCRIPTION_PROVIDER=mistral`
 - **Self-hosted**: Ensure ASR service is running on port 8765
 - Check transcription service connection in health endpoint
 

backends/advanced/README.md

@@ -31,7 +31,7 @@ Modern React-based web dashboard located in `./webui/` with:
 
 **The setup wizard guides you through:**
 - **Authentication**: Admin email/password setup with secure keys
-- **Transcription Provider**: Choose between Deepgram, Mistral, or Offline (Parakeet)
+- **Transcription Provider**: Choose between Deepgram or Offline (Parakeet)
 - **LLM Provider**: Choose between OpenAI (recommended) or Ollama for memory extraction
 - **Memory Provider**: Choose between Friend-Lite Native or OpenMemory MCP
 - **Optional Services**: Speaker Recognition, network configuration

backends/advanced/SETUP_SCRIPTS.md

@@ -15,7 +15,7 @@ This document explains the different setup scripts available in Friend-Lite and
 
 ### What it does:
 - ✅ **Authentication Setup**: Admin email/password with secure key generation
-- ✅ **Transcription Provider Selection**: Choose between Deepgram, Mistral, or Offline (Parakeet)
+- ✅ **Transcription Provider Selection**: Choose between Deepgram or Offline (Parakeet)
 - ✅ **LLM Provider Configuration**: Choose between OpenAI (recommended) or Ollama
 - ✅ **Memory Provider Setup**: Choose between Friend-Lite Native or OpenMemory MCP
 - ✅ **API Key Collection**: Prompts for required keys with helpful links to obtain them
@@ -43,10 +43,9 @@ Admin password (min 8 chars): ********
 -------------------------------
 Choose your transcription provider:
   1) Deepgram (recommended - high quality, requires API key)
-  2) Mistral (Voxtral models - requires API key)
-  3) Offline (Parakeet ASR - requires GPU, runs locally)
-  4) None (skip transcription setup)
-Enter choice (1-4) [1]: 1
+  2) Offline (Parakeet ASR - requires GPU, runs locally)
+  3) None (skip transcription setup)
+Enter choice (1-3) [1]: 1
 
 Get your API key from: https://console.deepgram.com/
 Deepgram API key: dg_xxxxxxxxxxxxx

backends/advanced/src/advanced_omi_backend/models/conversation.py

@@ -21,7 +21,6 @@ class Conversation(Document):
     class TranscriptProvider(str, Enum):
         """Supported transcription providers."""
         DEEPGRAM = "deepgram"
-        MISTRAL = "mistral"
         PARAKEET = "parakeet"
         SPEECH_DETECTION = "speech_detection"  # Legacy value
         UNKNOWN = "unknown"  # Fallback value
@@ -63,7 +62,7 @@ class TranscriptVersion(BaseModel):
         transcript: Optional[str] = Field(None, description="Full transcript text")
         segments: List["Conversation.SpeakerSegment"] = Field(default_factory=list, description="Speaker segments")
         provider: Optional["Conversation.TranscriptProvider"] = Field(None, description="Transcription provider used")
-        model: Optional[str] = Field(None, description="Model used (e.g., nova-3, voxtral-mini-2507)")
+        model: Optional[str] = Field(None, description="Model used (e.g., nova-3, parakeet)")
         created_at: datetime = Field(description="When this version was created")
         processing_time_seconds: Optional[float] = Field(None, description="Time taken to process")
         metadata: Dict[str, Any] = Field(default_factory=dict, description="Additional provider-specific metadata")

backends/advanced/src/advanced_omi_backend/services/audio_stream/producer.py

@@ -52,7 +52,7 @@ async def init_session(
             user_id: User identifier
             client_id: Client identifier
             mode: Processing mode (streaming/batch)
-            provider: Transcription provider ("deepgram", "mistral", etc.)
+            provider: Transcription provider ("deepgram", "parakeet", etc.)
         """
         # Client-specific stream naming (one stream per client for isolation)
         stream_name = f"audio:stream:{client_id}"

backends/advanced/src/advanced_omi_backend/services/transcription/base.py

@@ -36,7 +36,6 @@ class TranscriptionProvider(Enum):
     """Available transcription providers for audio stream routing."""
     DEEPGRAM = "deepgram"
     PARAKEET = "parakeet"
-    MISTRAL = "mistral"
 
 
 class BaseTranscriptionProvider(abc.ABC):

backends/advanced/tests/test_conversation_models.py

@@ -134,7 +134,7 @@ def test_add_transcript_version(self):
             version_id="v2",
             transcript="Updated transcript",
             segments=segments,
-            provider=TranscriptProvider.MISTRAL,
+            provider=TranscriptProvider.PARAKEET,
             set_as_active=False
         )
 
@@ -170,7 +170,7 @@ def test_set_active_versions(self):
         segments2 = [SpeakerSegment(start=0.0, end=5.0, text="Version 2", speaker="Speaker A")]
 
         conversation.add_transcript_version("v1", "Transcript 1", segments1, TranscriptProvider.DEEPGRAM)
-        conversation.add_transcript_version("v2", "Transcript 2", segments2, TranscriptProvider.MISTRAL, set_as_active=False)
+        conversation.add_transcript_version("v2", "Transcript 2", segments2, TranscriptProvider.PARAKEET, set_as_active=False)
 
         # Should be v1 active
         assert conversation.active_transcript_version == "v1"
@@ -213,7 +213,6 @@ def test_provider_enums(self):
         """Test that provider enums work correctly."""
         # Test TranscriptProvider enum
         assert TranscriptProvider.DEEPGRAM == "deepgram"
-        assert TranscriptProvider.MISTRAL == "mistral"
         assert TranscriptProvider.PARAKEET == "parakeet"
 
         # Test MemoryProvider enum

config.env.template

@@ -65,16 +65,12 @@ OPENAI_API_KEY = sk-xxxxx
 # SPEECH-TO-TEXT CONFIGURATION
 # ========================================
 
-# Primary transcription provider: deepgram, mistral, or parakeet
+# Primary transcription provider: deepgram or parakeet
 TRANSCRIPTION_PROVIDER = deepgram
 
 # Deepgram configuration
 DEEPGRAM_API_KEY = 90xxxxxx
 
-# Mistral configuration (when TRANSCRIPTION_PROVIDER=mistral)
-MISTRAL_API_KEY = 
-MISTRAL_MODEL = voxtral-mini-2507
-
 # Parakeet ASR configuration (when TRANSCRIPTION_PROVIDER=parakeet)
 PARAKEET_ASR_URL = http://host.docker.internal:8767
 

tests/configs/README.md

@@ -60,7 +60,7 @@ done
 
 When creating a new test configuration:
 
-1. **Name it descriptively**: `{stt}-{llm}.yml` (e.g., `mistral-openai.yml`)
+1. **Name it descriptively**: `{stt}-{llm}.yml` (e.g., `deepgram-openai.yml`)
 2. **Use environment variables**: Always use `${VAR:-default}` pattern for secrets
 3. **Set appropriate defaults**: Update the `defaults:` section to match your provider combo
 4. **Include only required models**: Don't include models that aren't used
@@ -124,7 +124,7 @@ Test configs use environment variable substitution to avoid hardcoding secrets:
 
 As you add support for new providers, create corresponding test configs:
 
-- `mistral-openai.yml` - Mistral Voxtral STT + OpenAI LLM
+- `deepgram-openai.yml` - Deepgram STT + OpenAI LLM
 - `deepgram-ollama.yml` - Deepgram STT + Local Ollama LLM
 - `parakeet-openai.yml` - Local Parakeet STT + OpenAI LLM
 - etc.