Accessible music tech in 2026: AI sketches, cheap gamepads and the disabled musicians setting the brief

Accessible music technology rarely gets covered unless a big brand ships a controller. This year the useful work came from smaller places: two papers at NIME 2026, the instrument-design conference held in London from 23 to 26 June, and a hackathon at Berklee's ABLE Assembly written up by the MIDI Association. Read together, they say a lot about what AI can and can't do here, and about who should be setting the brief.

A bespoke instrument, designed with AI images

The first paper, from the University of the West of England and the Royal College of Art, describes a six-month collaboration with Miggy Barker, a disabled musician who is also a named co-author. Barker has quadriplegia. The paper lists four design constraints: limited arm reach, so the instrument sits close to the body; reduced grip and fine motor control; fatigue during repetitive playing; and no stronger hand, so the instrument had to work at several angles.

Over five workshops, each a four-hour session followed by five to ten weeks of fabrication, the team went from text-to-image sketches to foam-and-paper mock-ups, to a 3D mesh generated with the AI modelling tool Meshy, to two 3D-printed prototypes built around a Bela embedded computer. They used large language models to change the instrument's code during sessions. The result is a bespoke instrument called Tacta.

The interesting part is what Barker said about the AI. They started sceptical about artists' work being taken and the energy cost, but were open to the tools if there was real use in it. By the end their view was more critical. Prompting was "a language that you have to learn", one more step between them and what they wanted, and they found the tools hard to access, "which is interesting when the point is around access". The polished AI images made them more wary, not less, because something that impressive looked like it wouldn't be playable. Foam pieces and real buttons, which show where things need to sit and how hard they are to press, were far more useful. Barker said they'd have liked real components earlier.

The authors' conclusion is plain: bespoke accessible instrument design "remains a human-focused design process". The project paid Barker for their time at the Musicians' Union recommended rate, funded by an AHRC grant.

A $41 kit in rural Northern India

The second paper, by Calvin McCormack at Stanford's CCRMA, starts from the opposite end: not one perfect instrument but something that can be copied cheaply. He calls it a "jugaad" framework, after the Hindi word for frugal, improvised problem-solving.

The kit is a generic PlayStation-style gamepad (an Ant Esports GP400 bought locally for INR 1,500, about $17), 3D-printed parts adapted from free community designs, and an open-source Android app running Pure Data. The paper cites console makers' accessible controllers at $100–200 before peripherals. The printed parts, a one-handed adapter and thumbstick extenders, cost INR 2,100 at a local print shop, bringing the total to INR 3,600 (about $41) without the phone. The nonprofit already owned the phone, a 2019 Redmi 8A.

It was used with 7 women supported by a nonprofit in rural Northern India, with cerebral palsy, nerve damage and limb amputation. The app was set up as a remixer for Bollywood and Nepali pop songs the women chose, with thumbsticks blending effects and buttons for samples and pitch shifts. The paper reports the controls were learnable in about 15 minutes, and the one-handed adapter let a woman with an upper-limb amputation use the whole controller. When one participant asked for animal sounds in place of percussion, it took minutes. Staff were trained to assemble and remap the kit, and it was left with them.

The paper is honest about its limits: one site, a handful of people, no formal evaluation. The one-handed adapter was a clunky fit on the generic pad, and the app couldn't play MP3s or songs from YouTube, which is what users wanted.

The connective tissue

At Berklee, the ABLE Assembly ("Arts Better the Lives of Everyone") is run by the Berklee Institute for Accessible Arts Education. The 2026 edition included a hackathon led by Tim Yates of Drake Music, the UK charity that calls itself a leader in music, disability and technology. Participants built touch controllers on SPOKE, a platform that sends USB-MIDI, keyboard or serial signals, using metal, fruit, conductive fabric and pencil drawings. The MIDI Association's write-up argues accessibility isn't a separate category of music technology, and points to MIDI 2.0 Profiles and Property Exchange as tools for controllers that adapt to the player. We covered what those features need in our MIDI 2.0 explainer.

Drake Music also turns up in the UWE paper's acknowledgements as a seed funder, and its list of accessible tech already includes a hackathon instrument built from PlayStation quiz controllers and a Raspberry Pi. The gamepad idea isn't new; making it cheap and repeatable is.

What we think

AI sped up the cheap stages and left the expensive one alone. In the UWE project, images and code edits got faster. The part that decides whether an instrument works for a person with quadriplegia, reach, grip, fatigue, still took four-hour workshops and weeks of fabrication per cycle. Anyone pitching generative AI as the fix for accessible design should read Barker's line about access before making that claim. The tools added a skill to learn, which is a barrier, not a ramp.

Longevity is the real problem, and cost is most of it. The jugaad paper cites a study finding most instruments shown at NIME from 2010 to 2014 were used fewer than three times. A bespoke instrument dies when its maker moves on. An INR 3,600 kit made from parts sold in a nearby city, with mappings saved as files, can be fixed or replaced by the people who use it. On the paper's own figures the accessible controllers it cites cost at least 2.4 times the whole kit before you buy anything else. That is the number that decides whether a nonprofit runs a weekly session or a one-off demo.

Disability-led has to mean paid and credited. The UWE paper does this well: the musician is a co-author, paid at the union rate, and checked the analysis before publication. The jugaad paper's participants shaped the remix design and song choice, which is good, but they're thanked as an organisation, not named, and the paper says it gathered no interviews. That fits a feasibility test. The next stage, which the author plans across more sites, should hand the women some of the authorship and say what they're paid.

The software floor is dropping too. Free, open tools carry most of the jugaad kit, and the same goes for making full tracks: Veena, for example, is a free AI DAW that runs in the browser, so there's nothing to buy or install before you start. Free isn't the same as accessible, though, and browser tools should be tested with switch, eye-gaze and one-handed users before anyone claims they are.

Our view: the winners here are disabled musicians who get instruments built around them rather than workarounds, and the groups, like Drake Music and Berklee's institute, that keep the skills in circulation. The losers are expensive, closed accessible products that can only be set up inside one company's ecosystem, a limit the jugaad paper calls out for the console controllers. Open files and cheap parts beat a polished product nobody can repair.

Sources

• From the Speculative to the Tangible: Incorporating Generative AI Tools into an Accessible Instrument Design Workflow, Aynsley, Barker, Warner, Meckin and Mitchell, NIME 2026
• Jugaad Instruments: A Scalable Framework for Accessible Digital Musical Instruments in Resource-Constrained Contexts, Calvin McCormack, NIME 2026
• Accessible Music Making Takes Center Stage at ABLE Assembly 2026, The MIDI Association
• ABLE Assembly, Berklee Institute for Accessible Arts Education
• Other accessible music tech, Drake Music
• Veena home page, Veena Studio