Meta Launches Music Generating AI That Can Produce Tunes From Text Inputs

Words to Music

Generative AI tools that can produce text and visuals are becoming widespread and being shared rapidly online. But what about audio and music made by AI?

We've seen some AI generated music already. A recent Drake and The Weeknd song was fully AI produced with no involvement from the artists, showing future music industry disruption. But generative AI can also help create new music.

Meta's new generative AI model, MusicGen, uses text or melody prompts to generate original music based on samples in its system. Essentially, you tell MusicGen what kind of track you want or hum a tune and it comes up with variations.

Check some of the music samples generated by MusicGen

MusicGen was trained on 20,000 hours of music including full tracks and instrument samples, giving it inputs for AI creations. It's not publicly available yet but could significantly change how musicians, marketers and others make music.

Like all AI creations, MusicGen could face legal challenges from the powerful music industry which is pushing to stop unauthorized use of copyrighted content, including sampling. This could eventually lead to new rules limiting systems like MusicGen.

In short, we may soon hear AI generated songs climbing the charts without realizing it. Though MusicGen's potential goes beyond copying, opening up new ways for many people to create music for different purposes.

Whether that future is desirable or not, it's coming and will give more avenues for music creation.

Read more about it here