Okay so im an engineer looking for some advice and am interested as to what you guys might say. Lets say i have thousands of calls and transcripts. What is the best way search for model that I can train, I am looking for something that can be self hosted with proven low latency on real time transcription
#Best outcome for voice to text model training
red nexus
rocky fossil
Why train from scratch? What is wrong with Whisper?
red nexus
Why train from scratch? What is wrong with Whisper?
Would you suggest instead self hosting whisper and using the data to fine tune?
i thought whisper was not build for live transcription. I don’t mind modifying it but i’m trying to get the latency as low as possible on live VTT
so i was thinking of training a model thats more compatible instead