Wyoming Pocket-TTS

Simple Wyoming TTS server using Pocket-TTS.

Steps

Run container
- Run: docker run --rm -p 10300:10300 ghcr.io/jw-200/wyoming-pocket-tts:0.0.4
Use custom voices (optional) (see more)
- Put .safetensors files in a local folder (example: ./custom_voices)
- Run with volume mount: docker run --rm -p 10300:10300 -e VOICE_DIR=/app/custom_voices -v ${PWD}/custom_voices:/app/custom_voices ghcr.io/jw-200/wyoming-pocket-tts:0.0.4
- Voice name format is: (Custom) <file_name_without_extension>
Connect from Home Assistant
- Add Wyoming integration
- Host: your Docker host IP
- Port: 10300

WYOMING_HOST (default: 0.0.0.0) - Bind address inside the container.
WYOMING_PORT (default: 10300) - Wyoming TCP port to expose.
DEFAULT_VOICE (default: (Predefined) Alba (en)) - Voice used when the client does not request one. (find other voices)
VOICE_DIR (default: /app/custom_voices/) - Folder scanned for custom .safetensors voices.
PRELOAD_VOICES (default: same as DEFAULT_VOICE) - Comma-separated voice names to preload at startup.
VOLUME (default: 1.0) - Output gain multiplier.
LOG_LEVEL (default: INFO) - Logging verbosity (DEBUG, INFO, WARNING, ERROR).
ZEROCONF (default: pocket-tts) - Zeroconf service name shown to Home Assistant discovery.