1010Benja has put a generative AI music tool in the credits, more or less, and the result is harder to dismiss than the usual pasteurized sludge that comes out of Suno.
The track is “Semiramis’ Dream,” the opener from his latest EP, Time Has Nothing To Do With What You Choose…. According to 1010Benja, the song is marked as AI-generated on Deezer because Suno was part of the production chain. He told Hearing Things that he and collaborator Patrick Holder started with Holder’s beat and Benja’s recorded vocals, then repeatedly pushed material through Suno, edited the results in Ableton, and sent revised versions back into the generator.
That workflow matters because it is not the usual prompt-and-post routine. Benja told Hearing Things that he wrote the lyrics, sang the vocals, worked with collaborators on the music, selected outputs, edited them, added layers, and ran the process again. He estimated he did this a few hundred times for “Semiramis’ Dream.” That is still AI generation. It is also a lot closer to production than vending-machine songwriting.
The Verge’s Terrence O’Brien, who reviewed the track, described the song as driven by a jungle beat with Benja’s clearly human voice at the center. O’Brien wrote that the most obvious AI element is a group of backing vocals, but said the processing sits inside a hyperpop-adjacent style where chopped, artificial-sounding vocals are already part of the genre furniture.
Benja is not presenting the Suno work as a secret. “We’re not hiding it,” he told Hearing Things. “We don’t feel any shame about it.” He also said he used generative AI because the sound he wanted was beyond what he could afford, joking that he had not hired a professional children’s choir. In the same interview, he said he had not paid rent the previous month and would not pay rent that month either.
The ethics are still ugly. Benja acknowledged to Hearing Things that generative AI carries environmental costs, and he framed the broader fight against corporate AI deployment in bleak terms: “The good guy doesn’t win this one,” he said.
There is also a practical question buried under the tech argument: whether Suno was necessary for this result. O’Brien argued that much of the sound of “Semiramis’ Dream,” including the choir-like vocals, could have been approximated with layered human performances and vocal processing. That is a fair jab. Pop production has been making voices sound synthetic for decades without needing a model trained on other people’s music.
The track arrives after 1010Benja’s 2024 release Ten Total, which earned Best New Music from Pitchfork. O’Brien wrote that the other three tracks on the new EP do not match that earlier record, while calling “Semiramis’ Dream” the standout.
One song does not redeem a platform. Suno remains best known for making music that sounds assembled from genre averages and algorithmic guesses. Benja’s track shows a narrower point: when a musician treats the generator as one ugly tool in a larger studio process, the output can carry a human fingerprint. That should make the AI music debate more annoying, which is usually what happens when the facts get involved.
This story draws on original reporting from The Verge.
