Almost exactly a year ago, I presented a few selected programs for composing music with AI in a blog (in German: How powerful are AI-based programs for composing music already?), ranging from www.soundraw.io to www.riffusion.com. In particular, the AI-based apps for creating music using text prompts were still in the experimental stage. But now there’s an update on the current development of composition programs.
Try it out for yourself – in fact, the programs presented here are (so far) free of charge; registration is uncomplicated, for example with a Google/Apple or X account. Both companies presented here were founded by former Google researchers.
Let’s start with www.suno.ai: On Soundcloud (CLICK HERE) you will find an example that gives you an idea of what you can achieve with a proper text prompt. And on the suno-Webseite you’ll find a ton of other examples in the menu item “Explore”.
In an interview with music magazine Rolling Stone the co-founder of Suno, Mikey Shulman explained that music is incredibly complex and difficult to reproduce compared to text and images. “It’s a wave. It’s a continuous signal.” . On sunoy, the individual text and the title are generated by ChatGPT.
The second app you should check out: www.udio.com. Udio is supported by prominent investors such as Andreessen Horowitz and artists such as Will. i. am and Common.
The quality of both Udio and Suno is stunning. However, some users have reported that their Udio recordings sound a bit clearer, especially when it comes to vocals. Overall, it becomes clear that in the not too distant future AI-composed music will play a role on streaming platforms such as Spotify or Soundcloud – on use case: playlists for background music.
Addendum from 01.07.2024
The world’s largest music labels are suing AI music generators: Universal Music Group, Warner Music Group and Sony Music Entertainment have filed lawsuits against Udio and Suno. They accuse the music start-ups of illegally training their AI models with protected material.
Led by the Recording Industry Association of America (RIAA), the labels are demanding up to USD 150,000 per work and potential billions in damages. According to the RIAA, the platforms are said to have imitated specific artists and hit songs. The companies are said to have already admitted to training with copyrighted material.
Just recently, both startups have received significant funding – Suno got 125 mUSD in May, Udio secured 10 mUSD.
While text and image generation platforms have faced legal challenges, AI music generators are now also coming under scrutiny. Interestingly, music labels are particularly active here and are looking for collaborations with AI companies. It remains to be seen whether the current lawsuits will set groundbreaking precedents and / or lead to heavy fines and new licensing models.
Source: From the AI Newsletter by Henry Hasselbach from 27.06.2024
