Ah ok. I have done that as well. Deepspeech and Speechbrain and the other open source models for transcription are unfortunately not good. Probably because they

Ah ok. I have done that as well. Deepspeech and Speechbrain and the other open source models for transcription are unfortunately not good. Probably because they don't have enough training data relative to the big guys. You should show CLIP - probably the best open source model I have seen as it was trained on a huge corpus.