Load these custom whisper models

yum-food / TaSTT

A free self-hosted STT for VRChat

MIT License

76 stars 4 forks source link

Load these custom whisper models #5

Closed fire closed 11 months ago

fire commented 1 year ago

Start by looking at the WhisperModel instantiation in transcribe.py. I'm using the faster_whisper library for inference. They may or may not have a way to load these checkpoints.

Most likely we would to fetch them from hugging face, then point WhisperModel at that file path.

I can also take a look at this, this is intriguing.

fire commented 11 months ago

Closing for now.