Microsoft AI Releases Impressive Transcription and Voice Models

Microsoft AI announced the availability of its latest MAI-Transcribe and MAI-Voice AI models for building voice agents.

“MAI-Transcribe and MAI-Voice models [offer] accurate, fast, low cost, and chart-topping audio understanding and generation for building the best conversational voice agents,” the Microsoft AI announcement explains. “We’re also announcing two new voice models: MAI‑Voice‑2.1 and our blazing-fast variant, MAI‑Voice‑2.1‑Flash. Together, they give users the fast and fluid building blocks to create conversational experiences, with no compromise on accuracy or voice quality.”

Microsoft AI says that its new MAI-Transcribe-2-Streaming model is top-rated for both final and partial transcripts on Artificial Analysis, and that it delivers low-latency, real-time transcripts in 60 languages with continuous language detection.

“Rather than waiting for someone to finish speaking before returning text, it produces its first hypotheses (known as ‘partials’) in just over 100 ms of receiving audio,” Microsoft AI says. “It then revises them as more context rolls in and commits a stable transcript right away. These partials enable voice-enabled applications to act on speech before the speaker even finishes … our internal evaluations show that words appear in the transcript 2x faster than with our closest competitor.”

MAI-Voice-2.1 is Microsoft AI’s strongest multilingual text-to-speech model yet, and it offers support for 23 languages and 26 locales while enabling a single ‘voice’ across all languages with a native accent. That is, instead of using one accent across all languages, it can switch on the fly as the language changes. And while MAI-Voice-2.1-Flash supports the same languages and cross-language speakers as MAI-Voice-2.1, it’s optimized for high-volume, latency-sensitive workloads and can generate 45 seconds of audio with an end-to-end latency of just 150 ms. Microsoft AI says that MAI-Transcribe-2-Streaming is a good choice for those building natural, low-latency voice agent experiences.

MAI-Transcribe-2-Streaming with MAI-Voice-2.1-Flash can also work together to keep conversations moving naturally whether they’re used for customer service agents, multilingual assistants, or interactive learning and media. You can experience them in the new Chatter demo on the MAI Playground.