Microsoft has announced the launch of its first in-house AI prototypes, reflecting its drive to strengthen its autonomy in this fast-growing sector.

The first model is called MAI-Voice-1, and is dedicated to generating natural sounds, while the second model is called MAI-1-preview, and is classified as a text-based model developed and fully trained in-house.

According to the company's statement, the MAI-Voice-1 model is capable of generating a full minute of sound in less than one second using just one graphics processing unit (GPU).

The model is already used in some Copilot Labs services such as Copilot Daily, which provides a daily audio summary of the news, as well as producing podcast-like discussions to illustrate topics, and users can try it out on Copilot Labs with the ability to adjust the tone of voice and the style of delivery.

The MAI-1-preview script model was trained using about 15,000 Nvidia H100 chips and is designed to handle text instructions and provide useful responses to everyday queries.