What's new¶
- Audiohook connection duration minimization. Now the connection is paused/closed as soon as the target audio amount has been collected.
- AI model is now configurable through new parameter,
model_hash
. - Replacement of
websocket
byTwilio
terms in theaudio_protocol
parameter for Twilio Media Streams connections. - New optional mechanism to fetch the collected audio for subject registration. The websocket API adds the new parameter
return_audio
, whilst the REST API provides a dedicated new endpoint for that same purpose.