We use GPT Realtime 1.5 for some of our single prompt voice agents. What model generates the text transcript?
Hey Mark, I just wanted to clarify, is the ASR coming straight from OpenAI (whisper)?
Hey @cole
No; Retell’s own ASR pipeline:
-
English: primarily Deepgram
-
Other languages: Azure Speech or Soniox, depending on the language
Again this is for producing the text transcript.