This Could Be The Algorithm That Frees Us From AI-Generated Music

I don’t think that musicians are inherently against AI-generated music when used as a tool, but they’re certainly insulted and outraged when it’s used instead of real human-generated music. This is especially egregious when a song generated by an algorithm is collecting Spotify or TikTok royalties while real artist’s tracks go unplayed. The real problem is that sometimes these AI-generated songs can sound so real that no one call tell the difference. But what if there was a foolproof way to spot a track generated by AI? YouTuber Benn Jordan thinks he’s found it.

Benn, who hosts a wonderful YouTube music and science channel, claims that he’s found a way to spot AI-generated music that so far has identified songs generated using the Suno platform (which many think is the best at creating human-like music) 100% of the time.

Taking it one step further, Benn scraped 560 of Suno’s top generated songs, and then discovered that out of those 560, only 11 were not being monetized on a digital streaming platform. The problem here is that the other 549 songs were being posted by someone posing as a real artist, and in some cases the songs received hundreds of thousands of plays.

In case you haven’t noticed, this has been a big deal lately, with many accusing Spotify of playlisting AI-generated music tracks intentionally so that don’t have to pay royalties.

Perception Is The Key

While Benn doesn’t go into exactly how his algorithm works, he does say that it has to do with the difference between lossy (like MP3s or any streaming service) and lossless (like a CD) audio. The AI platforms have all been trained on data-compressed audio, which can throw away as much as 80% of the audio information when it’s encoded, although most listeners can perceive the loss. Benn intimates that an algorithm that can tell the difference between the two is the way to detect the AI content. He says that he’s not revealing the secret so that the AI platforms don’t use the info as a workaround.

While I applaud him for working so hard on this project, the fact of the matter is that it won’t mean anything unless the streaming platforms adopt the algorithm to crack down on AI-generated music as it’s uploaded to the platform, and enforced their “no-Ai” rule for their catalog. That said, Benn has a huge following and one can only hope that the Spotifys of the world will be proactive in adopting this method, although Benn said he’s going to approach the Tunecores of the world about this first.

Without that adoption, it’s only a pretty cool parlor trick.