As generative AI tools continue to evolve New ways of creating content are emerging, including text, images, video, and music as well.
This is an area where companies like OpenAI or even Google are making their mark.using tools like Jukebox or MusicLM, respectively. Meta is also in this “race” and has recently unveiled a new tool called MusicGen.
MusicGen allows you to turn text descriptions into short music clips, only a few seconds long. Descriptions can also be accompanied by additional pieces of music, which the tool will use as a reference during the creation process.
According to the company, The system behind MusicGen has been trained on 20,000 hours of music. Also included in the process are 10,000 high-quality music tracks, as well as 390,000 music tracks from platforms such as ShutterStock and Pond5.
The Meta project is open source and anyone interested in learning more can visit the MusicGen page on github. A demo of the tool is also available and you can try it out with this website.
“Friendly zombie fanatic. Analyst. Coffee buff. Professional music specialist. Communicator.”
