Thursday, September 28, 2023

Listen to an AI sing an eerily human rendition of ‘Jolene’

Must read

Shreya Christina
Shreya has been with for 3 years, writing copy for client websites, blog posts, EDMs and other mediums to engage readers and encourage action. By collaborating with clients, our SEO manager and the wider team, Shreya seeks to understand an audience before creating memorable, persuasive copy.

AI-powered image generators have been getting most of the press lately. But musical machine learning models have quietly made great strides in recent years. Holly Herndon has been at the forefront of that revolution. She developed (with partner Mat Dryhurst) Spawn, a singing neural network, for her latest album prototype and released holly+ (in cooperation with Sounds never heard before) to the public last year, allowing anyone to use a model of Holly’s voice. Now she has released a new single, with the only vocals being her digital twin brother.

This cover of Dolly Parton’s “Jolene” plays it right on first hearing. Yes, it’s slower and quieter, but Ryan Norris, who handles the instrumentation, doesn’t take extreme liberties with the arrangement or sound palette. It simply exchanges delirious despair with plaintive resignation.

What makes it so captivating is that every vocal sound, down to the sharp inhalations before the harmonies kick in, was generated by Holly+. (That’s right, it “breathes”.) There’s not a human in sight of a voting booth here. Some of the phrasing is a bit pompous and there are the occasional digital artifacts, if you listen carefully. But overall, this digital model of Holly Herndon’s real voice is impressive in its ability to imitate its creator.

Until now, most major artists’ experiments with AI have focused on creating generative soundscapes or synth melodies. This is (to my knowledge) the first time a machine learning model has taken the microphone solo in a pop song.

Herndon already viewed the track in March Sonar Festival, but it largely flew under the radar until it received a proper release this week. (BTW: check out Sonar’s presentation for some really wild real-time demonstrations of the technology from Holly+ and Never Before Heard Sounds.) You can try to recreate the performance above by recording your own performance of Jolene and uploading it to Holly+ , but don’t expect the same reliability of the results through the web app. While it’s certainly a fun diversion, artists who are serious about using AI to improve their craft should find out spawnan organization launched by Herdon and Dryhurst earlier this year.

All products recommended by cafe-madrid have been selected by our editorial team, independent of our parent company. Some of our stories contain affiliate links. If you buy something through one of these links, we may earn an affiliate commission. All prices are correct at time of publication.

More articles

Latest article