Please add https://www.speechmatics.com/ in the transcription. It provides better than deepgram quality transcription, for non English, though with longer processing time or latency.

Also, when I recorded elsewhere and wanna put this on Hedy, I always ended up using local whisper or nemotron even though already selected deepgram. For a quicker work around I ended up upload my audio recording to deepgram to get less than 10s of transcription for longer than 1 hour of meeting/appointment recording, then copy paste the transcription to Hedy session. I expect to get integrated workflow when deepgram API was preselected.