3 ms·
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
by verdverm 2mo ago
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
- jeffbee 2mo agoI 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.
- verdverm 2mo agoIt's typical to use a special STT model (audio in only), which will not be able to clean up afterwards. If you are using an LLM for the STT part, you're leaving stuff on the table
- coder543 2mo agoI haven't tried it, but this looked promising for that exact task: https://huggingface.co/superwhisper/s1-mini https://huggingface.co/superwhisper/s1-mini