In Transcript Generate mode, it would be great if users could also choose speech recognition models from providers like Google or OpenAI to perform transcription tasks.
hi, thanks for the feedback! We’ve reviewed dozens of SST models and providers and at the moment only Whisper V3 Turbo (on a custom inference solution) offers the right combination of speed, cost, WER and output quality.
If a better model becomes available, we’ll simply switch to it instead of complicating the API with more choices.
Log in to comment and vote
Comments1
Rafal
Sep 16, 2025
hi, thanks for the feedback!
We’ve reviewed dozens of SST models and providers and at the moment only Whisper V3 Turbo (on a custom inference solution) offers the right combination of speed, cost, WER and output quality.
If a better model becomes available, we’ll simply switch to it instead of complicating the API with more choices.