Google’s intelligence that helps you compose a song starting from a whistled tune
Google’s intelligence that helps you compose a song starting from a whistled tune.
Google has created an AI that can generate pieces of music of variable length starting from a text request, or from a whistled tune.
Google researchers have created an AI that can generate pieces of music of variable length starting from a text request (prompt). Even more interesting, artificial intelligence can generate highly articulated songs starting from a melody, for example, whistled by the user.
Google’s project is called MusicLM and is yet another milestone in the field of generative artificial intelligence. If DALL-E is able to create images starting from the user’s instructions, while ChatGPT is able to process texts, MusicLM occupies a completely new scenario: that of musical composition .
The user can give the AI rather vague instructions, or he can provide extremely descriptive and detailed props. You can ask MusicLM to start the piece with specific musical instruments, recreating an atmosphere or keeping a certain rhythm. Then you can ask to continue with a different style and so on, indicating the time interval of each part.
An example of how AI works comes to us from the following prompt:
- electronic music played in a video game (0:00-0.15)
- meditation music played near a river (0:15-0:30)
- fire (0:30-0:45)
- fireworks (0:45-0:60)
The prompt in question is not particularly detailed or sophisticated, but that was enough for MusicLM to generate this short piece.
MusicLM can also theoretically generate human voices, such as a choir, but The Verge argues that this is where artificial intelligence is currently struggling the most.