All caps, no capping.
This is a demo using transformers.js and Whisper to generate caption files. It's meant as a proof of concept so a lot of features are missing and it's rough around the edges:
- doesn't use webworkers which makes the UI lockup during processing
- is mostly unstyled so the UI might be confusing
- no way to edit time chunks
- only uses small whisper model, which means a lot of errors
Some things it does do:
- 🏡 Fully clientside thanks to transformers.js
- 📝 Export to VTT and SRT formats.
- ✏️ Edit the transcription live
- Move transcription code to webworkers
- Allow users to select larger whisper models
- Add timestamp editing and the ability to add/remove text groups
- Translation to other languages?
- Add tailwind or something?