faster-whisper
A local speech-to-text server for Home Assistant's voice assistant: it turns what you say into text on your own server, so no audio leaves your network. It runs faster-whisper, a fast implementation of OpenAI's Whisper models, behind the Wyoming protocol that Home Assistant speaks. This template uses the CPU image; the image's gpu tag is for NVIDIA GPUs only and is not used here. The model you choose is downloaded the first time it starts, which can take a few minutes, and is kept in the app's folder. To connect it, open Home Assistant, go to Settings, Devices & services, add the Wyoming Protocol integration and enter this server's address and the Wyoming port (10300 by default). Then choose it as the speech-to-text service in your voice assistant. The server has no login, so keep it on your local network.
- Documentation
- https://docs.linuxserver.io/images/docker-faster-whisper/
- Template revision
- 2