Google has crossed a new threshold in AI-supported music production. The company’s Magenta Project, which it has been developing for a while, is expanding with Magenta RealTime, a new open source model that enables live and real-time music production.

Google will make live music with AI

Magenta RT, developed with DeepMind support, was introduced with a structure where users can directly intervene in the music production process. The system not only produces sound, but also offers an interactive music creation area with instant improvisations during performance.

Shiftdelete.net

Magenta RealTime is an autoregressive transformer model with 800 million parameters trained on approximately 190 thousand hours of stock instrumental music. The model uses a block-based structure that performs music production without delay.

It creates sounds in two-second chunks. This technique enables AI-supported improvisations by instantly responding to user prompts during performance. The model currently runs on the free Colab TPU infrastructure offered by Google.

Unlike previous models, Magenta RT does not only produce sound; It creates new sound blocks by analyzing previous sound outputs and style data provided by the user. Thus, users can instantly change musical styles, instruments and general sound structure.

The system features SpectroStream-based 48kHz stereo sound production. This technology replaces the previously used SoundStream. At the same time, the model works with a new embedded model called MusicCoCa. This infrastructure allows users to make more detailed guidance with both text and audio inputs.

Another striking feature of the model is that it turns live music production into an area of ​​discovery. Users can switch between two different musical styles, try out unusual instrument combinations and create new sound textures using their own samples.

Magenta RealTime opens up a creative playground in musical terms, while also offering an open source infrastructure for developers, artists and researchers. So what do you think about this? You can share your opinions with us in the comments section below.