Configuration Reference

All configuration is handled via environment variables. Create a .env file or pass flags to Docker.

License Configuration

LICENSE_KEY=your_key  # Leave empty for Developer Mode (4 concurrent calls)

Core Settings

# Server Ports
HTTP_PORT=8081
AUDIOSOCKET_PORT=8080
WEBSOCKET_AUDIO_PORT=8082

# Operation Mode
GATEWAY_MODE=false  # Set to true for OpenAI/ElevenLabs direct

Provider Selection

# Auto-select best provider based on language
STT_PROVIDER=auto

# Or manual selection
TTS_PROVIDER=deepgram
LLM_PROVIDER=n8n

API Keys (Required)

# Add keys for providers you intend to use
SARVAM_API_KEY=...
GROQ_API_KEY=...
DEEPGRAM_API_KEY=...
OPENAI_API_KEY=...
ELEVENLABS_API_KEY=...

Performance Tuning

Voice Activity Detection (VAD)

VAD_POSITIVE_THRESHOLD=0.5
VAD_NEGATIVE_THRESHOLD=0.35
VAD_REDEMPTION_FRAMES=8  # 768ms pause tolerance

Audio Buffering

MIN_SPEECH_DURATION=500
MAX_BUFFER_DURATION=5000

Advanced

# Enable TTS Caching (Recommended)
TTS_CACHE_ENABLED=true
TTS_CACHE_MAX_SIZE_MB=5000

# Enable Barge-in
ENABLE_BARGE_IN=true
BARGE_IN_THRESHOLD=500