Voice agents need a speech layer built for dialogue. While Twilio ConversationRelay handles the complex orchestration of audio streams, most teams traditionally stitch together ASR, voice activity detection, and turn-taking logic to approximate natural conversation. This standard approach can add latency and false interruptions, creating challenges for a seamless user experience. In this session, Deepgram's Nick Kaimakis walks through Flux, the first conversational speech recognition (CSR) model purpose-built for real-time voice agents. See how embedded turn-taking simplifies voice agent architecture, how Flux Multilingual handles 10 languages with real-time mid-call switching, and how to leverage the flexibility of ConversationRelay to configure Flux as your STT provider.

This session is sponsored by Deepgram