12:52
11:39
13:16
09:59
14:15
10:28
12:52
11:39
13:16
09:59
14:15
10:28
12:52
11:39
13:16
09:59
14:15
10:28
12:52
11:39
13:16
09:59
14:15
10:28
Developed by Meta's Audiocraft team, MusicGen uses artificial intelligence capabilities to create music based on text descriptions, similar to the musical equivalent of ChatGPT.
With MusicGen, users can enter a short description of the desired music style and click the "Generate" button. Within a short period of time, the artificial intelligence algorithm creates a 12-second track corresponding to the given instructions.
For example, a user can request "lofi slow BPM electro chill song with organic samples," and MusicGen will create audio similar to the popular Lofi Girl radio on YouTube.
MusicGen allows users to guide the AI by uploading their own songs, enabling the model to gain a better understanding of musical structure. Developer Felix Kreuk shared examples on Twitter, showcasing MusicGen's ability:
Currently accessible to the public on Meta's Hugging Face website, MusicGen comes with a limitation: it only generates instrumentals and lacks vocal capabilities.