Meta has developed a generative AI that can create music based on text input from its users. The AI, called “MusicGen,” can generate music completely from scratch based on the text it is given, such as “80s driving pop song with heavy drums and synth pads in the background,” or it can apply the command to an existing music piece.
You can test MusicGen here, and it will generate approximately fifteen seconds of music based on the text input provided by the user. Meta has also released it as open source, which means that with some knowledge, users can experiment with MusicGen and create a tool that better suits their individual preferences.
MusicGen has been trained by analyzing 20,000 hours of music, including 10,000 licensed songs, as well as 390,000 instrumental tracks from Shutterstock and Pond5, according to TechCrunch. In the embedded tweet, you can listen to two pieces that MusicGen created by applying a text command to an existing music piece.
