There’s a lot of buzz about music made by artificial intelligence. Critics brand it as dull, uninspiring and lacking artistic flair. I beg to differ. AI can generate incredible music in seconds, across any genre. And it’s only getting better and faster. However, the true symphony lies in how we orchestrate this technology.
Now, anyone with a dream and a computer keyboard can compose music just by describing what they want. Platforms like Suno and Udio are leading the development of generative audio. Both are accessible from easy-to-use web pages, and the audio quality is astonishing. You can pen your lyrics and have them transformed into a rock opera in minutes. Or perhaps you fancy an electronic dance anthem instead. The ease of song creation is set to cause an online musical explosion. Imagine an all-you-can-eat ice cream buffet, except the ice cream is made in a distant factory rather than your local artisanal spot.
Some critics say that AI only takes bits of actual songs, and the generation is not creative. This might be true, but many real musicians also work this way — they are influenced by other songs and artists.
Maybe the difference is that computers make accurate copies, and humans are limited to their skills.
We also make mistakes and beautiful imperfections that can lead to better music.
Certain AI software even gives the users full copyright over AI-generated songs, all for a nominal fee. This opens up endless possibilities: Imagine launching a radio channel that streams copyright-free lo-fi music 24/7 or crafting the perfect atmospheric background scores for your crime-themed podcast series. The question is: How does this impact the traditional catalog music industry?
The issue of permission
The generative AI music software is trained with a significant portion of music in every genre. Now, there are signs that, in some cases, this data was used without the permission of the musicians or the record companies. Some big labels now demand that generative AI audio companies stop the unauthorized use of content in training AI models. This sounds only fair because we live in a civilized world, and it would be polite to ask for permission or even compensate for using other people’s work.
A Significant portion of music producers already incorporate AI-powered tools at some stage in their production process. Many of the most used tools are aimed at helping in the mixing and mastering processes. But there are also new powerful tools to manipulate audio files. With AI-based stem separation tools it is now possible to isolate vocals or even just the bass line from any track. There are also tools that change the vocalist’s timbre to mimic any famous singer. The only real limits to these creative possibilities are imagination and copyright laws.
There are also signs that AI songs have included almost identical clips of actual songs when generating music this way.
Hundreds of front-row musicians have pledged against the “predatory” use of AI in music because, with these tools, it is possible to create an unlimited amount of music in the style of any major act, such as The Beatles or Billie Eilish. While in most programs, you can’t ask the AI to make you a song in the style of The Beatles, you can ask for a “’60s British rock”-style and get a very familiar sound. There are also signs that AI songs have included almost identical clips of actual songs when generating music this way. If these new creations are used widely commercially, what would be a fair compensation model for the artists who created the original material?
AI music will get better very quickly and soon, it will make things previously unimaginable. What if you like Jimi Hendrix and you could listen to different live versions of ”Fire,” with Hendrix playing a different guitar solo or arrangement each time? The AI could also create covers of new pop songs with that same distinct live sound.
Radio channels are facing a challenge because of copyright and ethical concerns. The internet will have fun with this new tech without fearing implications or sanctions. Anonymous new musicians can generate remixes of classical Sibelius music where Britney Spears is singing with lyrics from Nickelback. We will also hear more parody songs and other music memes. Do we need to find ways to keep up or make more regulations?
AI music is here to stay. Luckily, we still need people to play it live.
The author is executive producer for Yle in Finland.
This article was taken from a RedTech special edition, “Radio Futures: Keeping an eye on AI” which you can read here.
More music – stories that can carry a tune
MPW announces Women in Music Tech Summit
