Google introduced its new artificial intelligence (AI) generation of audio files through text descriptions. The name of this tool is MusicLM, and it has been shown a lot of usage forms for its advertisement.

For its operation, the system is trained to store 280,000 hours of ringtones, from there it generates the content that someone requests based on the phrase, situation and description of many options. “MusicLM is a model that creates high-fidelity music from text descriptions, such as “a soothing violin melody backed by a distorted guitar riff,” its developers explain on the AI ​​official page.

But in addition to creating compositional songs, he also has the ability to create melodic pieces by whistling and humming according to the rhythm that is required.

The potential of this artificial intelligence

The announcement of this new tool is accompanied by some examples of its use, assuming that the limitations are something that anyone can use due to the variety of uses.

The first parts of the ideas are directly related to music, such as the creation of an audio piece that combines different genres such as reggaeton and electronica and includes “spatial and otherworldly sound”.

But it’s also used to create an arcade video game soundtrack, giving it precise instructions such as “a fast, upbeat beat with a catchy electric guitar riff.”

Examples extend to more specific situations, such as running, meditating, waking up, or creating pieces to give 100% to a situation. Which expands the range of creativity to combine with other ideas.

One of the most striking points is the ability to describe works of art. The advertisement includes several examples of how a painting can be turned into music based on its description.

one of them Persistence of memory from Salvador Dali, who describes him as follows: “His images of melting clocks mock the rigors of chronometric time. The watches themselves are like soft cheese; In fact, according to Dali himself, they were inspired by hallucinations after eating Camembert cheese. In the center of the painting, under one of the clocks, there is a distorted human face in profile. Ants on a plate represent decay.”

And the result is calm music, as if you were on the beach in the middle of a trip, with soft melodies.

An example they repeat with such iconic works as Napoleon crossing the Alps by Jacques-Louis David dance Henri Matisse and Crying by Edvard Munch.

But the possibilities are endless as they generate audio that represents what a place sounds like, a musical genre chord, different decades or a musician’s experience.

Finally, they gave some examples of how the same request can produce different pieces, thanks to the large number of hours of music that this artificial intelligence has created.

At the moment this tool is not available for all users and therefore TechCrunchGoogle does not plan to launch it in the short term due to the challenges posed by copyright and copyright issues.

A large part of this was because the researchers found that at least 1% of the pieces generated were inspired by existing music.

“We found that only a small fraction of the examples were correctly remembered, and in 1% we identified a probable match.” We insist that we continue to work to address the risks associated with the generation of music in the future,” they said.