argmaxinc / WhisperKit

On-device Speech Recognition for Apple Silicon
https://takeargmax.com/blog/whisperkit
MIT License
3.17k stars 268 forks source link

Add `Progress` to `WhisperKit` #71

Closed finnvoor closed 6 months ago

finnvoor commented 6 months ago

Currently there doesn't seem to be a way to get the current transcription progress to display. This adds a Progress to WhisperKit for easily displaying transcription progress in a ProgressView as seen in the modified WhisperAX.

The progress could probably be much more fine grained by updating in the decoder loop (and possibly factoring in model load time/anything else?), but this is a pretty good start I think.

ZachNagengast commented 6 months ago

This is cool, good idea! Although it looks a bit strange in streaming mode because the full audio length increases with every loop, can we just disable it for the streaming tab?

finnvoor commented 6 months ago

This is cool, good idea! Although it looks a bit strange in streaming mode because the full audio length increases with every loop, can we just disable it for the streaming tab?

Didn't notice that, should be fixed now ✅