automatic reference text transcription

[cocktailpeanut]

Currently the voice cloning feature doesn't seem to work unless you provide an accurate transcription of the reference audio, which is too tedious.

This can be fixed by incorporating whisper to automatically transcribe the reference audio instead of making the users manually enter the reference text. This would make a huge difference since most people are too lazy to transcribe audio clips

[edwko]

Thanks for the suggestion, I’ll look into adding this.

[edwko]

Added in the 0.2.1 release :)