b6bd53b01a
- Update `.env.example` with comprehensive speech and AI configuration options - Modify Docker Compose speech configuration for more flexible audio and ASR settings - Enhance Dockerfile to support Python virtual environment and speech dependencies - Refactor environment loading to use Bun's file system utilities - Improve device listing tool with more detailed device statistics - Add support for multiple AI models and dynamic configuration
74 lines
2.1 KiB
YAML
74 lines
2.1 KiB
YAML
version: '3.8'
|
|
|
|
services:
|
|
homeassistant-mcp:
|
|
image: homeassistant-mcp:latest
|
|
environment:
|
|
# Speech Feature Flags
|
|
- ENABLE_SPEECH_FEATURES=${ENABLE_SPEECH_FEATURES:-true}
|
|
- ENABLE_WAKE_WORD=${ENABLE_WAKE_WORD:-true}
|
|
- ENABLE_SPEECH_TO_TEXT=${ENABLE_SPEECH_TO_TEXT:-true}
|
|
|
|
# Audio Configuration
|
|
- NOISE_THRESHOLD=${NOISE_THRESHOLD:-0.05}
|
|
- MIN_SPEECH_DURATION=${MIN_SPEECH_DURATION:-1.0}
|
|
- SILENCE_DURATION=${SILENCE_DURATION:-0.5}
|
|
- SAMPLE_RATE=${SAMPLE_RATE:-16000}
|
|
- CHANNELS=${CHANNELS:-1}
|
|
- CHUNK_SIZE=${CHUNK_SIZE:-1024}
|
|
- PULSE_SERVER=${PULSE_SERVER:-unix:/run/user/1000/pulse/native}
|
|
|
|
fast-whisper:
|
|
image: onerahmet/openai-whisper-asr-webservice:latest
|
|
volumes:
|
|
- whisper-models:/models
|
|
- audio-data:/audio
|
|
environment:
|
|
- ASR_MODEL=${WHISPER_MODEL_TYPE:-base}
|
|
- ASR_ENGINE=faster_whisper
|
|
- WHISPER_BEAM_SIZE=5
|
|
- COMPUTE_TYPE=float32
|
|
- LANGUAGE=en
|
|
ports:
|
|
- "9000:9000"
|
|
deploy:
|
|
resources:
|
|
limits:
|
|
cpus: '4.0'
|
|
memory: 2G
|
|
healthcheck:
|
|
test: [ "CMD", "curl", "-f", "http://localhost:9000/health" ]
|
|
interval: 30s
|
|
timeout: 10s
|
|
retries: 3
|
|
|
|
wake-word:
|
|
image: rhasspy/wyoming-openwakeword:latest
|
|
restart: unless-stopped
|
|
devices:
|
|
- /dev/snd:/dev/snd
|
|
volumes:
|
|
- /run/user/1000/pulse/native:/run/user/1000/pulse/native
|
|
environment:
|
|
- PULSE_SERVER=${PULSE_SERVER:-unix:/run/user/1000/pulse/native}
|
|
- PULSE_COOKIE=/run/user/1000/pulse/cookie
|
|
- PYTHONUNBUFFERED=1
|
|
- OPENWAKEWORD_MODEL=hey_jarvis
|
|
- OPENWAKEWORD_THRESHOLD=0.5
|
|
- MICROPHONE_COMMAND=arecord -D hw:0,0 -f S16_LE -c 1 -r 16000 -t raw
|
|
group_add:
|
|
- "${AUDIO_GID:-29}"
|
|
network_mode: host
|
|
privileged: true
|
|
entrypoint: >
|
|
/bin/bash -c " apt-get update && apt-get install -y pulseaudio alsa-utils && rm -rf /var/lib/apt/lists/* && /run.sh"
|
|
healthcheck:
|
|
test: [ "CMD-SHELL", "pactl info > /dev/null 2>&1 || exit 1" ]
|
|
interval: 30s
|
|
timeout: 10s
|
|
retries: 3
|
|
|
|
volumes:
|
|
whisper-models:
|
|
audio-data:
|