Comment by WinH

Yes. We run run Whisper Large V3 (not Turbo) for the speech to text. It still seems to be the best open source model out there for that step. The main challenge we are trying to solve is Speaker Identification, which is a very time consuming process.