Pickr is an award-winning Australian technology news, reviews, and analysis website built to make technology easier for everyone. Find the latest gadget reviews, news, and more focused on the only ad-free tech site in Australia.
How much of what you listen to was created without the aid of artificial intelligence? Depending on what you listen to, the answer might surprise you.
There’s no shortage of music, with a veritable smorgasbord of songs in so many styles, but there’s also a type of music that appears to be taking over: AI music.
Built with the help of artificial intelligence — either generated partially or entirely — AI music is music that has been generated by a computer from a prompt, and likely from one of the many AI music systems found around the web. Suno, Udio, and plenty of others have popped up, taking text prompts like the image generators, but spitting out audio in sections of 30 seconds and higher.
Fit enough sections together and you might just have a real song. Use the programming interface for these systems, however, and savvy individuals might be able to build automated music aplenty. Run a script and create music without much effort at all.
It doesn’t take much to bring these tracks to music services, either, thanks to distribution systems that do the work for you. No record label needed or even any fame, as we found out when we made our own AI album; simply pay a fee and press submit, and the AI music is online for all to listen to.
From there, would-be musicians can aim to collect an eventual royalty or two. Depending on how many tracks or albums they’ve submitted, that could result in a hefty payday at the end of each month.
Which may well be the point, at least with some of the clear growth in AI music.
Deezer has been talking about just how many AI tracks are being uploaded to its service, and if the number is as big here, it’s likely as sizeable on other services, too.
One of the rivals to Spotify and Apple Music, the provider has noted that it receives “over 20,000 fully AI-generated tracks on a daily basis”, essentially accounting for over 18 percent of all uploaded content, up from 10 percent back in January.
These are changes, for sure.
Helping get to that number is an AI detection tool Deezer launched at the same time, which gives it the ability to remove what it recognises as fully AI generated content, all thanks to algorithmic approaches.
Most of this appears to target the AI approaches used by the models from Suno and Udio, at least from the automatic sense. AI music with a bit more work thrown in, such as what we did stemming from our own AI music article, appears to have flown by under the radar (not that anyone is listening).
But it does paint an interesting picture about how much AI music is appearing on services, particularly since those same songs could be found on other services where an AI reduction tool may not be present.
“It’s a crazy world we’re living in,” sang Jamiroquai back in 1996 about technology before the world of AI really rose to a common existence. Back then, and even up until a few years ago, the music you listened to would have been mostly human. These days, it may as well carry a bit of a question mark.
At the moment, what you’re listening to every day may not be fully AI-engineered.
If you know the artist, it’s probably not an AI artist. Meanwhile if you don’t, what you’re listening to certainly could be AI. However, many of the tools used in music creation may mean regardless of what you listen to, elements of AI could be used.
While the tone-matching technology that is “auto-tune” isn’t technically intelligent, there are versions of auto-tune that are AI-based, not to mention AI instruments and short stem splitting in Logic Pro when musicians and engineers are building tracks. Mixing and mastering these songs can also be improved by degrees of AI, some of which can be software services specifically using AI for these purposes.
The point is that whether you’re listening to something partially AI generated or fully AI made, there’s a decent chance that unless the track was recorded with just a few people in a studio with instruments and a microphone, elements of AI may have touched the music creation process.
That is, of course, strikingly different to fully AI generated music, the likes of which Deezer’s system focuses on. It will be easier to generate electronic tracks in fully AI made ways, more so than say jazz, which could be complex by comparison.
But it also means there with nearly 20 percent of music additions arriving to services daily being AI generated in the first place, you now have a greater chance of listening to something built by a computer from a text prompt about an idea.
Whether that makes it any more or less interesting as a piece of music will come down to you, and how quickly you press fast forward.
Leigh 🙂 Stark is an award winning technology journalist with over 16 years of experience, having appeared at the Financial Review, Popular Science, GadgetGuy, and on TV, radio, print, and web.
AI music on the increase with more on streaming services – pickr.com.au
