Hello, I am leading a small team right now, and we are developing an AI call center with Twiliio, and we're using Deepgram both for STT transcription and TTS. We are integrating into businesses soon and have clients lined up. You guys are the best. We love your STT and your TTS. They're both awesome products.
Currently, we are using Twilio's Gather function with speech_model set to deepgram_nova-2. We also have barge_in set to True, which allows interruptions.
We are exploring the possibility of directly integrating with Deepgram STT API with webhook/streaming to reduce costs since Twilio charges us a large fee for using gather. Do you have any advice, tips, docs, or even just features that will assist us to implement interruption handling so that a customer could interrupt the Deepgram TTS, start listening (with Deepgram STT), and then automatically end point?
Thank you!