ElevenLabs Adds Direct Publishing of AI-Generated Music to SoundCloud
ElevenLabs now lets users publish AI-generated music made on its ElevenMusic platform directly to SoundCloud, according to TechCrunch, which reported the change on November 5, 2026.
The integration comes with a promotional offer. SoundCloud Artist Pro users are being offered three months of ElevenMusic Pro. That tier allows generating up to 1,000 songs, splitting song stem files 500 times, and creating five custom styles or vocals per month. Stem splitting separates a finished track into its component parts, such as vocals, drums, and bass.
ElevenLabs released its first music-generation model in 2025. In April 2026 it released a consumer app that generates songs from text prompts. A month later it released a new music model that can switch genres mid-track and handle more complex vocals. The company said users have created more than 25 million tracks using its models.
The SoundCloud feature is not the company's only route to distribution. ElevenLabs launched its Music Marketplace on Thursday, March 19, 2026, so that users could publish original tracks made with its AI technology, as Billboard reported at the time. Creators can list songs, and other users can purchase usage rights to use, remix, and download those songs for their own projects. According to ElevenLabs' documentation, one of the usage types covers distribution to streaming platforms, including Spotify, Apple Music, and SoundCloud.
The generator turns text into songs, instrumentals, and melodies, and can produce complete songs with vocals, according to the ElevenLabs music page.
For producers who already work in a digital audio workstation (DAW), the software used to record and arrange music, the practical change is in the workflow. Stem separation is now a standard step in remixing and sample-based production, so the 500-split cap matters to anyone building on generated material. The limit of five custom styles or vocals per month is tighter. It suggests the company is gating its heavier customization features behind usage tiers rather than offering them broadly.
In my view, the more consequential question is provenance. A track generated in one tool, published on a host with a large creator base, and then licensed through a marketplace to a third party passes through several hands. Each step raises questions about attribution, licensing terms, and what a listener is told about how a song was made. The feature announcement does not settle those questions. The platforms that distribute this material will need to answer them in their own policies.
The broader context here is a shift in who gets to make recorded music. Tools that turn a text prompt into a song lower the cost of entry for people without instruments, studio time, or a label contact. Some of that activity will be routine, some experimental, and some commercial. The useful test for the coming year is whether these channels give independent creators a clearer path to an audience, and whether listeners can tell what they are hearing.
