logoalt Hacker News

verdvermyesterday at 9:22 PM2 repliesview on HN

have another model do a pass to clean it up, saw a demo of local STT where someone did this, can fix a lot of things, especially with gotchas for the STT model in a clean-transcript.md


Replies

coder543yesterday at 9:43 PM

I haven't tried it, but this looked promising for that exact task: https://huggingface.co/superwhisper/s1-mini

jeffbeeyesterday at 9:40 PM

I think the model can even evaluate itself. If it looks afterward at an output like "do you. Want to get lunch?" in the absence of affirmative evidence that the user wanted it that way, it should be able to see that it goofed.