Startup Jen has unveiled an AI platform for music generation that combines high audio quality with strict copyright compliance.
Jen has introduced its music generation platform called Jen [ALPHA]. The company describes it as an "ethically trained" text-to-music system aiming to set a new standard for copyright compliance in AI-generated music.
According to Jen, the platform uses proprietary, high-resolution, waveform-based and omnidirectional diffusion models to generate high-quality stereo audio tracks using a novel neural network architecture.
Users can generate audio tracks up to 40 seconds long via text prompts, with options to extend them. The company says it has filed several patent applications for its technology.
Video: Jen
Jen claims it used over 40 fully licensed music catalogs for initial training. Each generated track is automatically checked against a database of 150 million tracks for audio recognition and copyright identification. This includes both the compositions in the training dataset and every new track created on the platform.
Additionally, Jen generates a cryptographic hash for each track, stored on the Root Network blockchain. This aims to prove the integrity and timestamp of each track's creation. Verified tracks receive a "JENUINE" label.
The company targets both beginners and professional producers. Users should own the tracks they create. Jen plans to introduce a feature allowing users to sell their creations.
"It's very simple. #dontstealmusic," the company summarizes its philosophy in a tweet.
Jen is positioning itself in a controversial area of AI development: while the technology opens up new creative possibilities, it also raises complex issues of copyright and fair compensation for artists.
Competitors like Udio and Suno have faced criticism for training text, image, and video AI models on artist data without consent. Unlike its competitors, Jen currently only offers instrumentals - vocals are not available.
