#voice AI

Mira Murati’s startup has introduced its first model, arguing that voice AI should process conversation as it happens instead of waiting for rigid turn-taking.

Wispr Flow says India has become its fastest-growing market after a Hinglish rollout, highlighting both the opportunity and the complexity of building voice AI for multilingual users.

Uber says it is using OpenAI models to power conversational assistants and voice features that help drivers interpret earnings opportunities and help riders complete bookings more quickly.

OpenAI has introduced three new audio models aimed at turning voice interfaces into more capable real-time systems that can reason, translate, and transcribe while conversations are happening.

OpenAI says developers can now build applications that converse, translate and transcribe in real time through new voice intelligence features in its API.

OpenAI says natural voice AI depends on low and stable media round-trip time, quick session startup, and global reach. A new engineering write-up shows how the company rebuilt key parts of its WebRTC stack to meet those

Google has begun rolling out Gemini 3.1 Flash TTS, a new text-to-speech model aimed at giving developers and enterprise users finer control over synthetic voices while extending support to more than 70 languages.

DeepL has launched a voice-to-voice translation suite and an API for custom deployments, expanding from text translation into meetings, mobile conversations, and frontline worker scenarios.