Ideas to correct Wav2Vec2 transcription results

Hi,

I think you need a language model trained on a large corpora. You can have a look on this issue.

Best,
Omar