Skip to content

BRK244 Breakout Thu 19 Nov 16:00 PST

Voice agents in Microsoft Foundry: Prototype to production in 45 mins

Yiwen Rong, Archer Zhao, Marvin Hsu, Matt Landis

Summary

From Microsoft's AI summary

Based on Microsoft's AI summary: Microsoft's AI speech team showed how to take voice agents from prototype to production in Foundry. The Voice Live API is stated as generally available, with flagship, mini and lite real-time models. New speech features include multilingual transcription with container support, HD voices, a VASA-powered photo avatar and Live Interpreter. Foundry covers build to monitoring, with agent handoff and built-in phone support through partners such as Twilio. No pricing, regional availability or GA status was stated for the new speech features.

Session description

Microsoft's catalogue description

Voice is the fastest-growing way people use agents, and the hardest to put in production. Voice-first agents in Microsoft Foundry give you one place to build a speech-native agent, ship it to web, mobile, Teams and telephony, and operate it with audio-aware tracing and evaluation. We'll cover what ships, how it differs from voice-enabling a text agent, and what enterprise customers are running today.

← All sessions