The Interledger Community 🌱

Discussion on: Automated captions for everyone with Waasabi

maboa profile image
Mark Boas

Hey Flaki - nice write-up!

If you have words, media but no timings you can align (for free) using something like Gentle (English only). Then convert to a Hyperaudio Lite (MIT Licensed) based Interactive Transcript using our converter - source code here.

Alternatively you could take a look at our Wordpress plugin.

flaki profile image
Flaki Author

Thanks Mark! I didn't know about Kaldi/Gentle, looks like a really cool tool to generate more precise timings for the transcripts to feed back/retrain our models based on the media content, thank you for the tip (and the praise).

Yes, the plethora of timed transcript formats I found while checking out Hyperaudio was also super useful, so we don't reinvent the wheel but use one of those to create our generated transcripts.