The company Suno, based in Massachusetts, has launched its third text-to-music conversion tool, which produces two-minute songs in any style and with vocals. And the worst part is that it’s really good.
Last year at this time, MusicLM, from Google, pointed out that mixing and production were next on the AI list making short clips of instrumental music where the whole band, orchestra or ensemble were generated together. And a whole series of IA startups were starting to have fun with voices, allowing people to digitally transform their voices into more famous ones, for example.
Suno belongs to a different group of AIs that simply do everything for you, all at once. Write what kind of song you would like to listen to and it will generate a couple of options for you in about 30 seconds.
Will AI put an end to human music?
You can give vague or very specific instructions. You can write your own lyrics or give an idea of how you would like it to sound. You can tell him what kind of instruments you would like to hear or let him surprise you with his style.
From time to time, some strange artifacts appear in the audio, it has a clear preference for self-tuned voices, and you can be sure that the lyrics it comes up with will be cheesy at a GPT level unless you give it some solid direction.
However, when it hits the nail on the head with a song, it becomes difficult to know that it is created with AI. It is surprising to see how many aspects of music it has absorbed, digested, and is capable of regurgitating. The future is terrifying.
