Pappo isn’t alive to read this but he has a heart attack anyway: Google has a project called MusicLM, an AI that composes from text. We tell you what it’s all about, below!
In recent times, lots of images generated by artificial intelligence, which were produced from text like Dall-e, were seen on networks. But now Google He assures that he has a project on the way that will do the same for music, and that it is called MusicLM. The idea is that you will be able to describe how you want a song to sound and this technology will do the same to give you a result that fits, style “Pink Floyd covering the killer bombshell“.
MusicLM uses some 28,000 hours of pre-recorded music content to generate the results, and it supposedly includes various genres, rhythms, and instruments. The issue is that, since it has not been officially launched at this time, it cannot be tested. However, they assure that you can go from reggaeton to dance and make music in the style of documentaries, video games, soundtracks and more.
Another virtue of this AI is that once a base is created it can be modified and new melodic lines can be added. However, the problem that academics ran into is that a high percentage of the sounds generated so far (1%) are too similar to already existing sounds. Added to this are problems imitating the human voice.
For these reasons, Google said that it will not release this AI to the public at this time. However, the project is already there for the future.
![](https://culturageek.com.ar/wp-content/uploads/2023/02/MusicLM-culturageek.com_.ar-1.jpg)
In the past, traditional musicians always criticized alternative ways of making music, and the case of guitarist Pappo telling DJ Dero that his thing was not music and that he should look for an honest job is famous. The situation was so funny that it became a meme. However, Google’s proposal goes a little further: what happens to music when it is not directly generated by a human?