Called “Project Music GenAI Control,” this experimental tool leverages AI to generate original compositions based on simple text descriptions, then gives users full control to refine the auto-generated tracks until they have customized music perfectly suited to their needs and tastes.
Everything you need to know:
✓ Adobe’s AI tool Project Music GenAI generates music from text
✓ Users can customize AI music by editing tempos, patterns, structure
✓ Tool aims to make music production easy for creators and amateurs
As someone who enjoys listening to podcasts in their free time but lacks the technical skills to produce their own themes and interludes, I was intrigued by Adobe’s vision for this new AI music creator.
In today’s blog post, I’ll unpack how Project Music GenAI Control works and what opportunities – as well as questions – it presents for both amateur and professional audio creators.
How does Adobe Project Music GenAI Control work?
Here’s the basic workflow:
↳ Users enter a short prompt like “upbeat dance music” or “calm nature track.”
Project Music GenAI Control’s AI model then auto-generates a musical composition styled to match that descriptor. Currently, the tool can produce music across genres like pop, hip hop, jazz and classical.
What’s the difference maker?
However, what really differentiates this prototype from existing AI music services is its integrated editing panel. Without leaving the app or requiring any formal training in digital audio workstation software, users get full pixel-level control over elements like tempo, melody patterns, intensity and structure. Sections can be rearranged, extended or remixed on the fly.
Project Music GenAI can generate loops. Great for video content like podcasts, tutorials & more.
Got a reference track? Project Music GenAI control generates a similar track
Users can also refine auto-generated tracks by providing a reference melody for the AI to emulate and potentially merge styles – like generating a hip hop song that incorporates elements of classical composition. This degree of customization allows shaping finished tunes to creative briefs or personal tastes.
Now, how was Adobe able to achieve this?
To train its deep neural networks, Adobe reportedly fed the system exclusively public domain compositions from historical greats like Beethoven and Billie Holiday. No privately authored works were involved. The aim was to teach the algorithms music fundamentals while avoiding potential legal issues around IP and sampling.
So in summary, Project Music GenAI Control streamlines music creation by offloading the initial composition to AI, then provides an intuitive interface for non-technical users to iterate on auto-generated results until achieving the desired final product.
AI editing may do for audio what pixels did for photos.
While the tool is still early-stage, its accessible editing capabilities could lower barriers for businesses, podcasters and other casual content creators seeking easy custom music solutions.
At the same time, the prototype suggests generative AI may eventually enhance – rather than replace – the roles of sound designers, music producers and composers. By quickly fleshing out initial motifs, these systems could fuel greater experimentation and refinement of audio drafts in professional workflows too.
Of course, many open questions remain.
“…[T]he tool gives creators deep control over how they can shape, adjust, and edit audio just like you control images at the pixel level.”
Nicholas Bryan, senior research scientist at Adobe Research, said in a press release
Will tech make music easy for amateurs?
One prototype says maybe. But what will that actual interface look like? How intuitive will modifications really be for lay users without digital audio experience? Only time and further iterations will show.
For now, the prototype demonstrates how generative AI is maturing to simplify specialized tasks across creative fields. Where the tool goes from here will be fascinating to observe – both for indie podcast enthusiasts seeking quick bed music, as well as audio professionals curious how new technologies may augment their roles. If Adobe’s vision comes to fruition, custom music production could become nearly as accessible as snapping a smartphone photo or tweeting a thought. Only continued development will determine if Project Music GenAI Control hits all the right notes.
