Skip to content

OpenVoiceOS/ovos-stt-plugin-vosk

Repository files navigation

Description

Mycroft STT plugin for Vosk

Install

pip install ovos-stt-plugin-vosk

You can download official models from alphacephei

Configuration

You need to download a kaldi model or provide a direct download url

  "stt": {
    "module": "ovos-stt-plugin-vosk",
    "ovos-stt-plugin-vosk": {
        "model": "/path/to/unzipped/model/folder"
    }
  }
 

Advanced configuration

  "stt": {
    "module": "ovos-stt-plugin-vosk-streaming",
    "ovos-stt-plugin-vosk-streaming": {
        "model": "http://alphacephei.com/vosk/models/vosk-model-small-en-us-0.15.zip",
        "verbose": true
    },
    "ovos-stt-plugin-vosk": {
        "model": "http://alphacephei.com/vosk/models/vosk-model-small-en-us-0.15.zip"
    }
  }
 

ovos-stt-plugin-vosk - regular STT, transcription happens after recording finishes

ovos-stt-plugin-vosk-streaming - streaming STT, transcription happens real time

verbose - print partial transcriptions

model - full path or direct download url for model

lang - optional, if model not provided will download default small model (if it exists)

Docker

This plugin can be used together with ovos-stt-http-server

docker run -p 8080:8080 ghcr.io/openvoiceos/vosk-stt-http-server:master