New York (EFE).- Google has created an artificial intelligence tool called MusicLM that can generate musical pieces lasting several minutes from text guidelines and, although the model is not open to the public at the moment, the company has already shared examples of what you can do.
In one example, the researchers called for: “A fusion of reggaeton and electronic dance music, with a spacey, otherworldly sound. It induces the experience of being lost in space, and the music would be designed to evoke a sense of wonder, while being danceable,” and MusicLM responded with a 30-second piece with beats, instruments, and melodies.
You can listen to the various samples here: https://google-research.github.io/seanet/musiclm/examples/
MusicLM is also capable of recreating human sounds such as a hum or whistle and even reproduce how an opera singer or choir would sing.
However, in the examples that the company has shared, the words that are sung are melodic, but do not correspond to any language.
In some cases the vocals and instruments sound a bit shrill, but in others it’s impossible to distinguish between the MusicLM versions and the human-made ones.
“Our experiments show that MusicLM outperforms previous systems in both audio quality and adherence to text description,” the company says in a statement.
The specialized media The Verge has indicated that the company does not plan to launch MusicLM at this time and that one of the reasons is the risks of misappropriation of creative content, as well as possible cultural appropriation or misrepresentation.