We have a new model artificial intelligencewho wants to revolutionize the world of music. Is named MusicLMhas been created by Googleand is capable of convert a text description in a melodyin any musical genre you ask.
There are already different AIs that compose music, for example Riffusion. But they are still quite primitive, they are not at the level of their “colleagues” who create drawings or photographs. This can change with MusicLMwhich according to the examples that Google has shown, does obtain coherent results and melodies that seem to be composed and performed by a human musician.
This artificial intelligence thing has caught Google on its feet, it has panicked when it has seen that people began to use ChatGPT to search for things, instead of their search engine. Many of the dismissals of a few days ago are aimed at “cleaning up”, to focus on AI.
He is already working on more than 20 artificial intelligence models, and as we are going to see, the effort is paying off.
MusicLM, Google’s AI that composes music
As Google itself explains in the press release, MusicLM is a model that generates hi-fi music from textual descriptions as “a relaxing violin melody accompanied by a distorted guitar riff”.
MusicLM generate 24 kHz music which remains constant for several minutes. It can condition both the text and a melody, since it can transform whistled and hummed melodies according to the style described in a text caption.
As in other AI models, the key is in the text entered. He acknowledges things like “the main soundtrack of an arcade video game. It’s fast and upbeat, with a catchy electric guitar riff. The music is repetitive and easy to remember, but with unexpected sounds, like cymbal bangs or drum rolls.”
Or also “A fusion of reggaeton and electronic dance music, with a spatial and otherworldly sound. It induces the experience of being lost in space, and the music would be designed to evoke a sense of wonder and wonder, while still be danceable”.
A novelty of this type of artificial intelligence is that You can merge different themes, one after the other, indicating the duration of each fragment. For example:
- Electronic song played in a video game (0:00-0:15)
- Meditation song played by a river (0:15-0:30)
- Fire (0:30-0:45)
- Fireworks (0:45-0:60)
It is also capable of putting music to pictures and images, describing what they show. Or ask for more generic things, like “elevator music,” or “improvised jazz.”
Another very practical function of this Google AI is put music to landscapes or situations. It is enough to say “a tropical beach with crystalline waters and palm trees that generate shade”, “a rainy and cold day seen from the window of a cafeteria”, or “a thief escaping from jail”. It is easy to imagine its application in movies, video games, commercials, etc.
Finally, MusicLM may create music with different skill levels. You can ask him to use beginner, intermediate, or professional interpreters.
All these descriptions are real examples that Google has used, and that you can listen to on this website.
MusicLM has been trained with 280,000 hours of music, including copyright-protected tunes and songs. That is the reason that Google does not dare to publish the AI for people to use, as they could receive complaints similar to those received by Stable Diffusion or MidJourney. First he wants the legal issue to be clarified.
After conquering the world of image and text, the artificial intelligence now he dares with musicthanks to MusicLM of Google. A new issue of conflict with musicians who live off compose melodies or arrangements for all kinds of projects.
George is Digismak’s reported cum editor with 13 years of experience in Journalism