AIs can generate not only text, images, and video, but also sound and music. The progress in quality is amazing.
Let’s look at three prominent examples:
Udio
Launched a week ago as part of a public beta, Udio has already caused quite a stir. The website contains numerous examples of songs created with this tool. You can currently try it out for free.
Ars Technica took a closer look and came to a familiar conclusion: yes, good results can be achieved, but only with effort, patience and a bit of luck.
As a user, you can have the lyrics generated automatically or you enter them yourself. You additionally describe which music genre you want. Two clips will then be created. Optionally, you can expand these music snippets in the next step.
Udio declares that it does not reserve any rights to the songs and that they can be used commercially.
Suno
When Udio showed up, the hype around AI music had already been started by Suno. Udio and Suno are very similar. Udio offers some additional options, such as expanding a music clip. However, testers found the quality of Suno’s songs to be better.
Rolling Stone magazine described it as “ChatGPT for music”.
Suno offers free access, which allows you to create ten songs per day.
Stable Audio
Stable Audio goes in a different direction. Here you can create three-minute songs that follow a typical structure. However, they do not contain any vocals. It is therefore suitable for instrumental pieces or as a starting point for your own production.
Stable Audio can not only create music based on a text prompt (text to audio), but can also use an existing clip as a starting point (audio to audio).
More examples
There are several other examples:
- Soundry AI is aimed at musicians and generates set pieces such as beats.
- Soundraw, in turn, advertises itself as a tool for both content creators and musicians.
- Beatoven.ai specializes in the target group of creators – think YouTube videos, podcasts, games, and audio books.
- Loudly from Berlin emphasizes royalty-free music for content as its benefit.
- AudioCraft is an offering from Facebook/Meta and can create both music and soundscapes.
- Boomys is apparently aimed primarily at people who have always wanted to become musicians.
- AIVA describes itself as an AI music generation assistant. The focus here is still on people.
- Hydra wants to appeal to both companies and musicians.
- Soundful advertises itself as a solution for royalty-free background music.
- Wavtool sees itself as an assistant for music producers.
- Amadeus Code is intended as a stimulus for new song ideas.
- Adobe calls its tool Project Music GenAI Control, which is designed to help musicians.
- Tuney also sees itself as a supporter for musicians and helps with generating, producing, editing and remixing.
I am sure this list is incomplete.
Conclusion
Ultimately, there seem to be three main audiences here. First, I see content creators, including companies that need music to accompany content such as videos or podcasts. Second, musicians who are interested in tools that make certain tasks easier or promise loops at the touch of a button. And third, amateurs who have always dreamed of making music.
Personally, I’m not sure how big the market for these products is. Time will tell!
By the way, many of the tools mentioned now disclose where they got their training material from. For professional use, it is also important to know if and under what conditions you are allowed to use your results commercially. Unfortunately, a look at the terms and conditions of the service is often the only way to find out.