Record, transcribe, TTS between callers

[Vieri]

Hi,

When a caller calls another one, is it possible to preprocess the caller’s and callee’s audio streams so they are “fed” to a transcribing ASR (vosk or whisper) and a TTS so that the latter resulting audio will replace both the caller’s and callee’s original audios?

Maybe with Stasis() but any ideas / pointers?

Thanks

UPDATE: the transcription and TTS would need to be on-going, a bit like Record() in an infinite loop with silence detection. On silence → run ASR + TTS and playback.

[jcolp]

ARI + Dialling + Bridges + External Media + Bidirectional Audio.

All the pieces are present to do it using ARI, but you have to put them together.