Recently, Microsoft's artificial intelligence department officially released its first self-developed AI model, named MAI-Voice-1 and MAI-1-preview. This marks Microsoft's further development in the field of artificial intelligence, especially in the competition with OpenAI.
MAI-Voice-1 is a voice model that can generate one minute of audio in less than a second, and it only requires one GPU to achieve this. Microsoft stated that MAI-Voice-1 has been applied to multiple functions, such as "Copilot Daily," which allows an AI host to read the day's top news for users and generate discussion similar to podcasts to help explain various topics.
Users can experience MAI-Voice-1 in Copilot Labs by inputting what they want the AI model to say and choosing different voices and speaking styles. In addition, Microsoft also launched the MAI-1-preview model, which was trained on approximately 15,000 Nvidia H100 GPUs, mainly targeting users who need models that can follow instructions and provide help with daily queries.
Mustafa Suleyman, Microsoft's Chief AI Officer, mentioned in an interview last year that the company's internal AI models do not focus on enterprise-level application cases. He emphasized that Microsoft is committed to creating products that are very useful for consumers and has rich predictive capabilities in advertising and consumer behavior data. In the future, MAI-1-preview will be applied to some text usage scenarios of the Copilot AI assistant, which currently still relies on OpenAI's large language model.
Microsoft stated in its blog: "We have ambitious plans for the future, not only pursuing further progress, but also believing that by coordinating a series of specialized models for different user intentions and use scenarios, we will unlock significant value."
Official blog: https://microsoft.ai/news/two-new-in-house-models/
Key points:
🌟 Microsoft has launched two self-developed AI models, MAI-Voice-1 and MAI-1-preview, enhancing its competitiveness against OpenAI.
🗣️ MAI-Voice-1 can quickly generate audio and has been applied to multiple functions such as Copilot Daily.
🚀 MAI-1-preview will be used for text processing in the Copilot AI assistant, marking a new development in Microsoft's consumer-level AI field.
