This is a stripped down version of whisper.cpp that only includes the encoder.
To build execute ./build
.
To run the executable (first see model file prep instructions below) do: encoder-cli model-file.pth audio-file.wav
The encoder-cli
executable returns a JSON-formatted string to stdout.
Note that the encoder will ignore audio files that are less than 1 second in duration.
This code also needs a stripped down Whisper checkpoint file with the decoder part removed and converted to the GGML format. To do this, you can use this Google Colab notebook: https://colab.research.google.com/drive/1NO_J8-7ZwziBvKsLOr6B9lI32ewRkxOU?usp=sharing