Home Tech New ‘ethical’ AI music generator can’t write a halfway decent song

New ‘ethical’ AI music generator can’t write a halfway decent song

0 comments
New 'ethical' AI music generator can't write a halfway decent song

It is true that our tests The artists pushed Jen beyond the boundaries of what a “normal” person might ask in a query, veering more toward a “record store clerk” level of familiarity with recorded sound. Cleveland, for example, failed to get anything good out of a query about “mid-tempo California garage rock influenced by 70s Indonesian pop,” while Heywood expressed dismay that Jen didn’t seem to acknowledge his request for “pop city,” a type of Japanese music that rose to prominence in the mid-1970s and has seen a minor resurgence in popularity in recent years. But for Heywood, that kind of musical breadth is necessary, especially as a musician.

“A lot of musicians or producers, when they ask each other for something, will use bands and other artists as a reference point, like, ‘Let’s go for a Prince-type sound,’ or, ‘Let’s add some Clavinet like Stevie Wonder,’” Heywood explains. Because Jen doesn’t understand the artists currently recording and even some fairly common genres and instruments, it’s hard to come up with anything specific.

“I kept trying to get some warmth, like a vinyl hiss or saturation or something lo-fi or vintage-sounding, but everything I did had the same kind of hi-fi, video game menu screen sound,” Heywood says. “They even give you ‘lo-fi’ as a suggestion, but that didn’t seem to have much impact. If you’re trying to get a certain sound, like ’80s funk, the closest you can get is something that sounds more like Daft Punk.”

Every electric guitar sound WIRED and its testers generated sounded almost too clean, and it was virtually impossible to get it to produce a track that wasn’t in 4/4 time unless you used the word “waltz” in the prompt.

Jen co-founder Shara Senderoff says some of this is to be expected. The tool is in its alpha phase, and the 10- and 45-second tracks it generates are “meant to inspire and provide a starting point for creativity, not necessarily a final product,” she says. New capabilities are on the way, and because Jen was trained using a limited data set, it has room to grow and “will expand significantly in the beta phase,” Senderoff adds.

Everything Jen did Heywood says rock music was akin to the “clip art version” of the genre. Cleveland was able to pull off some songs that sounded “like they could be used in a car commercial” or were “stepping into Black Keys territory,” but he says that more than anything, he felt all of Jen’s musical suggestions were just cheesy.

“It seemed like the kind of music I would make if I was joking around with my friends about clichés from other genres,” he says. “I could see some of the songs on a really bad Netflix dating show, but nothing I made felt personally threatening.”

But what about all those who make the songs heard on a Netflix dating show? Could Jen be a threat to their jobs? According to Blickle, almost certainly.

“If you’re a producer with a small budget and you’re just trying to get your content out there, you can now say, ‘I’m not even going to pay a designer or an animator. I can use an image generator,’” he says. “The same goes for a music budget. If they can’t afford to pay anything for something that was going to cost them $2,000, then great — someone will think that’s $2,000 in their pocket.”

You may also like