This fork provides recent docker builds. It also changes the English model from 0.22 to 0.42-gigaspeech.
This is a server for highly accurate offline speech recognition using Kaldi and Vosk-API.
There are four different servers which support four major communication protocols - MQTT, GRPC, WebRTC and Websocket
The server can be used locally to provide the speech recognition to smart home, PBX like freeswitch or asterisk. The server can also run as a backend for streaming speech recognition on the web, it can power chatbots, websites and telephony.
For documentation and instructions see Vosk Website