MAI-Voice-1: An advanced audio model
Microsoft has launched MAI-Voice-1, a high-definition, ultra-fast voice generation model that can produce a full minute of audio in less than a second using just one GPU. This model is used in features such as Copilot Daily to provide daily news summaries in a virtual voice, and is also used to create podcast-like dialogues to explain various topics, making the user experience more interactive and lively.
User Experience and Voice Customization
Users can try MAI-Voice-1 on the Copilot Labs platform, where text can be entered and the style and tone of voice can be customized to suit the desired purpose. This customization allows users to produce personalized audio content suitable for a variety of uses, from news coverage to educational and entertainment content, enhancing the appeal of the content and increasing listener engagement.
MAI-1-preview: A comprehensive language model
In addition to audio, Microsoft introduced MAI-1-preview, a universal language model trained using about 15,000 Nvidia H100 GPUs. This model focuses on providing useful and efficient answers to everyday queries, while continuously improving the user experience. Microsoft has begun publicly testing this model on the LMArena platform to measure its performance before gradually integrating it into various Copilot features.
Developing AI in-house
This move shows Microsoft's commitment to developing AI technologies internally, minimizing reliance on external technologies such as OpenAI. However, the company still uses OpenAI technologies in some of its services, indicating continued collaboration between the two companies to ensure the best user experience while leveraging each party's capabilities.
Exploring and Experimenting with Models
For those interested in experimenting with these models or exploring how they can be used in different applications, Microsoft provides a Copilot Labs platform that allows users to interact with the voice and language models directly. This experience gives the user a chance to understand the advanced capabilities of the models and test their ability to fulfill everyday or professional needs in a seamless and efficient way.
Microsoft unveils MAI-Voice-1 for ultra-fast and accurate voice transmission.
