/silero

transcribe audio/video files to text

Primary LanguageRust

Silero

Specify input files in wav/weba/webm/opus/vorbis format and it will transcribe them to txt in an optional output directory.

Dependencies

  • libonnxruntime
  • libopus

AI model

Download the AI model using models/en/download.sh.

What to expect

  • installing libonnxruntime is a pain, but tract can't load the model currently
  • the matroska library is poor and will fail to parse many valid webm files
  • only english is currently supported

License

Apache-2.0 + MIT