Microsoft has unveiled its first in-house AI models: MAI-Voice-1 and MAI-1-preview. MAI-Voice-1 can generate a minute of audio in under a second on one GPU and already powers Copilot features like daily news recaps and podcast-style explanations. Users can test it on Copilot Labs with customizable voices and styles. MAI-1-preview, trained on 15,000 Nvidia H100 GPUs, is designed for consumer-focused text tasks and will soon roll out in Copilot. Microsoft is also testing it on the LMArena benchmark platform as part of plans to build a range of specialized AI models.
Source: The Verge
