The Moises SoundCloud partnership, announced on October 21, 2026, lets artists pull their own tracks from SoundCloud into Moises, split them into stems, mix and master them with AI, and publish the result straight back to SoundCloud. If you keep demos, bootlegs or half-finished mixes on SoundCloud, this is for you. Until now that meant downloading, re-uploading and renaming files until one of them was called “final_FINAL_v3.wav”. Below is what the deal includes, what it leaves vague, what both companies promise about AI training and what we would do with it this week.
The short version: your track leaves SoundCloud, gets taken apart, comes back louder and goes home. A round trip, with a mastering chain in the middle.
What happened so far
The deal makes more sense once you know SoundCloud’s recent history with AI. That history has been bumpy.
- January 25, 2024: SoundCloud announced AI integrations with Fadr (remixes), Soundful and Voice-Swap.
- February 2024: SoundCloud added a terms clause letting uploads “inform, train, develop or serve as input to” AI, per Music Business Worldwide.
- May 14, 2025: after a backlash, CEO Eliah Seton said the wording “was too broad” and promised opt-in consent before training generative models on an artist’s voice, music or likeness.
- October 21, 2026: Moises and SoundCloud announced the partnership from Salt Lake City, with Pro cross-trials.
- A day later, on October 22: Music Ally called the SoundCloud partnership with Moises a promotional deal with a technical integration.
- By October 25: Mixing.co.kr had reported the details.
How the Moises SoundCloud partnership works, step by step
The partnership starts inside Moises: you connect your SoundCloud account there, on mobile or on the web. Then you import “eligible” tracks. (Nobody has defined eligible yet. The word turns up in the announcement several times and never explains itself.)
From there the tools split into two jobs. For practice and learning you get stem separation, which means splitting a finished stereo mix into separate parts like vocals, drums, bass and the rest. You also get chord detection and lyric detection. So a guitarist can mute the guitar on their own old demo and play over it, which is more useful than it sounds.
For production, tracks open in Moises Studio, a browser-based audio workstation. There you can mix, master, “reimagine” a track with what Moises calls contextual stem generation, and invite producers or other artists to work on the session in real time. An assistant called Moi does the chores. It is “agentic”, meaning it carries out steps itself instead of only answering questions: it sets up sessions, mixes and masters to your instructions, and teaches.
Then the round trip closes. Finished tracks publish directly to SoundCloud, where eligible artists can monetize qualifying music. That’s “eligible” again, still undefined.
The commercial side of the Moises and SoundCloud partnership is simple. Moises Pro subscribers get two months of SoundCloud Artist Pro, and SoundCloud Artist Pro members get two months of Moises Pro.
The Moises SoundCloud partnership covers your own uploads, not everyone else’s
The headline makes it sound as if every track on SoundCloud is now a stem pack. It isn’t. Every description of the Moises SoundCloud partnership, from the companies’ own release to Mixing.co.kr, talks about tracks from your own account. Nothing in the announcement says you can split someone else’s song.
That matters on a platform where bootlegs and unofficial remixes have always been half the culture. If you separate another artist’s vocal (with any tool) and publish a remix, the rights question hasn’t moved an inch. If anything, monetizing it makes the question louder.
Rights holders still check uploads, too. Mubert’s own data, last 28 days: about one in four copyright checker results found a rights holder’s claim on the uploaded track (25.7%, down slightly from 27.1%). A tidy master doesn’t make a borrowed vocal yours.
The AI training promises, and why SoundCloud had to make one
Moises and SoundCloud both put their training rules into the partnership announcement, which tells you they expected the question. SoundCloud says it does not train generative AI on artists’ music without explicit consent, and that its approach rests on consent, attribution, transparency and creator control. Moises says it trains only on licensed or authorized music, never on what users upload or create.
SoundCloud’s line has a history behind it. The 2024 clause and its 2025 rewrite are the reason “explicit consent” now opens every SoundCloud AI sentence. A partner that announced anything less would have walked into the same fire.
Moises also says it focuses on supporting artists’ ideas rather than generating complete songs. Fair enough. Still, “reimagine tracks through stem generation” means the software is writing new parts, so the line between assisting and generating is drawn in pencil. We saw the same pencil when Mozart AI swore off finished songs and then started finishing them.
The 150 million artists figure needs a calculator
The press release headline promises “150+ million artists”. The same release says Moises is used by 80 million musicians and SoundCloud hosts more than 40 million artists. That adds up to roughly 120 million, before you subtract the people who use both. The missing 30 million go unexplained.
The release also says Moises processes nearly six years of audio every day. That is longer than plenty of bands stay together, and some of them spend it arguing about one snare. Anyway.
What we think about the Moises SoundCloud partnership
The useful part of the partnership between Moises and SoundCloud is boring, which is a compliment. Removing the download, convert, upload, retitle loop saves real hours for anyone who uses SoundCloud as a sketchbook. Stem separation on your own old demos is also an underrated way to practise, and chord detection helps when you’ve forgotten what you played in 2019 (we have, more than once).
The strategic part is distribution. Moises gets a publishing button with an audience behind it, and SoundCloud gets a production tool that feeds its catalogue and its paid tier. That’s why Music Ally calls it promotional before it calls it technical. Editing inside finished audio is quickly becoming a standard feature everywhere, which is also why Apple bought Samplab.
One confession about the AI mastering. We once approved a master on laptop speakers at 3 a.m., and on every other system it sounded like a toaster with a law degree. Automated mastering is a good first draft. Check it on headphones, a car and a phone before it goes public.
What to do with the Moises SoundCloud partnership this week
- Connect a test account first and see which of your tracks count as “eligible”.
- Run stem separation on an old demo and practise over your own part.
- Use the cross-trial if you already pay for one Pro plan, and set a reminder before month two ends.
- Keep bootlegs of other artists’ songs out of the publish button unless you have permission.
- Compare the AI master with your own on three different speakers before you replace anything.
- Read the monetization terms on SoundCloud before you assume a remix earns money.
The track goes out, gets taken apart and comes home louder. Just make sure it was yours when it left.
Who we are
