SourceAudio and Musical AI Ink Strategic Partnership Adding Output Attribution To Most Comprehensive AI Music Dataset Marketplace

25 June 2025 - Press release

 

First and leading service for ethical generative AI music licensing and first true attribution tech provider for AI join forces to support growing demand for fairly trained AI models. 

SourceAudio and Musical AI are announcing a strategic partnership that will unite their AI expertise and provide licensed, high-quality datasets for AI training and output attribution on an opt-in basis. The companies will use their opt-in catalogs and proprietary technology to allow AI companies to train models on fully licensed data, then, if desired, identify the source of AI outputs and pay rightsholders accordingly. 

“We are very aware of the growing needs of both large platforms that are household names and newly founded startups that want to ensure the data they train their models on is fully licensed. This approach is the only truly reliable way to create generative AI products that acknowledge and reward human creativity,” said Drew Silverstein, President, Head of AI at SourceAudio. “By teaming with Musical AI for output attribution services, we are providing a game-changing marketplace for labels, publishers, and independent artists to connect with generative AI companies and work together, all while opening up new revenue opportunities on the input and now output side.”

As part of this collaboration, SourceAudio clients will have the option to participate in joint SourceAudio-Musical AI datasets. Musical AI will ensure these tracks are made available for potential training in rightsholder-determined ways and that any outputs generated from rightsholders’ works are attributed to the proper parties. This way, generative AI companies can accurately and fairly pay rightsholders for use of their work. 

This partnership will allow both companies to empower and support major players in tech and entertainment looking to pursue the proper approach to human-created works: to license them. “We are here to make sure our future customers can do the right thing by licensing and paying for their use of copyrighted music in training,” says Sean Power, CEO and co-founder of Musical AI. “SourceAudio’s outstanding catalog and community will play a vital role in our mission to ensure all generative AI for music is fully licensed and fairly trained.”

SourceAudio’s AI music dataset licensing marketplace launched in May 2025 and has already generated over $1.35 million in new annual recurring revenue for the rightsholders who opted in. Currently, the marketplace offers over 14 million fully cleared tracks, including nearly 4 million with vocals, as well 3 million sound effects and 200 sampled instruments - with more tracks from independent labels, publishers, and artists being added daily. Several AI companies, including Wondera.ai, are already onboard and using these songs to power their generative AI engines ethically. All songs are fully cleared on both sides – recording and composition – and are available along with standard and enriched metadata, MIDI, song segment analysis, ground truth and/or AI stems, trend data, and real-time usage information to power predictive capabilities for all types of AI-enabled music products.