"The company says its new MAI-Voice-1 speech model can generate a minute's worth of audio in under one second on just one GPU, while MAI-1-preview "offers a glimpse of future offerings inside Copilot.""
""So, we have vast amounts of very predictive and very useful data on the ad side, on consumer telemetry, and so on. My focus is on building models that really work for the consumer companion.""
""We have big ambitions for where we go next," Microsoft writes in the blog post. "Not only will we pursue further advances here, but we believe that orchestrating a range of specialized models serving different user intents and use cases will unlock imm"
Microsoft introduced two homegrown models: MAI-Voice-1 for fast speech synthesis and MAI-1-preview for instruction-following text tasks. MAI-Voice-1 can produce a minute of audio in under one second on a single GPU and is already used in Copilot Daily and for generating podcast-style explanatory discussions. Copilot Labs enables users to enter text and customize voice and speaking style. MAI-1-preview was trained on roughly 15,000 Nvidia H100 GPUs and targets helpful everyday responses. MAI-1-preview is being tested on LMArena and will be rolled into Copilot for select text use cases.
Read at The Verge
Unable to calculate read time
Collection
[
|
...
]