We’ve written a lot about AI music creation and how initial and understandable pangs of concern from human songwriters – these robots are stealing my job! – has faded as the realisation that AI music creation tools will mostly be in the guise of songwriting creative partners.
Music-making platform Splice – which says it has paid out tens of millions of dollars to creators – has launched a new AI music creation platform that helps further flesh out our ideas of how human-AI creative partnerships will work. It’s in the form of an easy-to-use sampler/synth app called “CoSo”, meaning “complimentary sounds”.
That’s a useful way to think about how music is created in the app: users are supplied with a constant supply of loops that can fit together and be combined in different styles, keys, and tempos to make fuller songs. These loops of music can then be dragged into a DAW and shaped into a wider piece of work – or they can be uploaded to TikTok, for users to weave into UGC projects.
So what will this kind of easy-to-use, swipeable music creation technology mean for creators? “We will hear a richer and more diverse selection of sounds bubbling up through this technology, that we might surprise creatives with new ideas,” says Ale Koretzky, Splice’s machine learning and audio science innovation lead.
We’ve wondered a lot recently about how recorded music will fit into the “web3 generation” of creator-consumers: people who don’t expect to merely be passive consumers of music (or other content), but to be able to take it apart and rebuild new versions of it, in, for instance, a metaverse space. It’s likely that this UGC-native group will want music to be not only plug-and-play, without any complex licensing issues, but also as adaptive components that can be changed, or be contextual.
We spoke to Vickie Nauman about this on our podcast recently, where she explained the complexity of licensing existing big-name music for the metaverse (in her case, 20 David Guetta songs for a Roblox event).
So perhaps the music from AI-driven music creation platforms is the missing part to this puzzle, and a more realistic future of music in UGC/web3 experiences may be a mixture of high-quality and easily re-tooled AI music, plus the occasional well known, fully licensed song. It also suggests that our current idea of “music consumers” may expand rapidly: there is quite probably a lot of people who love music – and are willing to pay for it – but not in the traditional form of pre-recorded songs.
AI music creation might be an opportunity to grow the music pie – and the definition of what both a music fan and a creator is – as opposed to eating away either.