Piano Lab Pro: Amanda L. Cowell

Orientation video: what the student experiences inside the headset.

Piano Lab Pro is a mixed reality app on Apple Vision Pro that lets piano students practice in front of a virtual audience, in a virtual concert hall, before they ever set foot on the real stage.

I led the call for proposals, the evaluation framework that selected this project from the candidate pool, and the scoping conversations that narrowed three faculty concepts into one focused product. From there I owned product through discovery, framing, build, and validation, working with a core team of six and 35+ contributors across design, research, and production. Two faculty partners shaped the work from inside their disciplines: Aya Higuchi Hagelthorn, Lecturer of Piano Pedagogy, Director of Collegiate Class Piano, and Coordinator of the Piano Pedagogy Laboratory Program, brought the pedagogical anchor. Anıl Çamcı, Associate Professor and Director of Graduate Studies in the Department of Performing Arts Technology, led the spatial audio design.

As of May 2026, the project is on track for its June 2026 MVP. The load-bearing pieces are complete.

Problem

Music performance anxiety is well-documented, widespread, and structurally built into how musicians train. Students rehearse alone in small practice rooms, then perform in large concert halls in front of live audiences. Repetition in a practice room doesn't close that gap.

Piano students are particularly exposed. Their instrument is fixed, you can't carry your home environment with you. What you can do is make the practice environment more like the performance environment before the performance happens. That's the thesis behind Piano Lab Pro.

Across our research, the moment students named most consistently as the anxiety trigger wasn't the performance itself. It was the transition into it: the audience lights dim, the stage lights come up, the room goes quiet. That moment became the product's core design requirement, and it's what makes Piano Lab Pro different from a generic VR practice room.

Side-by-side: a piano classroom on the left, Hill Auditorium from the stage on the right — Where students practice. Where students perform. The product closes the gap between them.

Approach

Discovery to framing to inception to MVP. I led several months of research before development began so the team could scope with confidence.

Discovery. Several rounds of user research ran in parallel with site visits to Britton Recital Hall and Hill Auditorium. The methods varied (interviews, surveys, in-context observation), most of which I can't share publicly to protect student data.

Synthesis from student interviews surfaced four anxiety themes that clustered around the moment of transition into performance: fear of failure and judgement, preparedness and cognitive slips, responsibility and external expectations, and awareness of physical anxiety response. To pressure-test where those anxieties actually lived, we ran a co-design workshop in which students mapped venue type against audience type and marked where they felt most anxious. Two clusters emerged. The dominant one (14 dots) was large venue with strangers, which became Hill Auditorium in the product. The second (8 dots) was small venue with familiar audience, a different kind of anxiety that justified building Britton Recital Hall as a second environment.

Co-design workshop: students mapped venue size against audience familiarity and marked where they felt most anxious. Two clusters emerged: large venue with strangers (14 dots) and small venue with familiar audience (8 dots). — Two synthesis artifacts: where students said those anxieties actually lived, and the themes that emerged from interview synthesis.

Build. The student wears an Apple Vision Pro at a real piano. Through passthrough they see their hands and the real keys. Everything else is virtual: Hill Auditorium or Britton Recital Hall, a live audience, calibrated venue reverb. The flow is continuous. Default state has the house lights up and the audience chatting. When the student is ready, they trigger the transition: lights shift, crowd quiets, stage is lit. The "all eyes on me" moment is now part of the practice session.

We built Britton first on purpose. It was the smaller, more accessible venue, and it let us work out the capture-and-build workflow before committing to the higher-stakes Hill Auditorium session. By the time we walked into Hill, the process was tested: 6,000+ photos and 4,200 aligned in photogrammetry, and 20 balloons popped at mapped locations to capture acoustic impulse response. Audience capture meant 42 people filmed on green screen on the XR Stage, 2.7 TB of footage to populate virtual seats with real human motion. Doing Britton first made Hill possible in a single capture day instead of three.

Scheduling access across two auditoriums, a green screen stage, and SMTD piano classrooms, while coordinating faculty, facilities, and the XR team across a full semester, was the stakeholder management layer beneath every image on this page.

Britton Recital Hall scene in Unity: audience panels, spatial sound sources, optimized piano mesh

The app is one piece of the product. The other is the student guide, a scaffolding piece designed in collaboration with behavioral scientists. Students take a self-assessment that helps them identify their own anxiety triggers and surfaces specific strategies they can then practice inside the app. The guide isn't supplemental, it's how a session is meant to work: identify, prepare, practice, reflect. Building the app without it would have made the experience effective for some students and inert for others.

Piano Lab Pro Student Guide, a Google Site with preparation content deployed to students

Validation. Multiple rounds of user testing across the project, formal and informal, focused on environment realism, keyboard alignment, and instruction clarity. The findings drove product decisions, not just polish. I also brought in U-M's Institute for Social Research to design the formal impact study, because self-reported anxiety scores weren't going to be enough on their own to validate whether the app worked.

Decisions

Three decisions worth surfacing, out of many that shaped this project.

01

Built the evaluation framework that selected the project

When I joined CAI, the team was about to receive a fresh round of XR proposals with no shared structure for evaluating them. I built the framework from scratch before any proposals came in: criteria spanning learner outcomes, technology fit, and operational fit, plus a free-form section for the things that don't score cleanly. Every team member rated independently. After scoring, we gathered red flags, identified what was worth saving in each proposal, and met with faculty to refine the ideas before any final call.

On the criteria alone, Piano Lab Pro wasn't the clear winner. But that wasn't the framework's job. The framework's job was to surface what to ask faculty about, and the scoped-down version of Piano Lab Pro that emerged from those conversations was the clear choice.

What I learned. A first-pass framework should be deliberately broader than what you'll end up using. The redundant criteria are the ones that teach you which criteria actually matter; you can't subtract before you've added.

02

Scoped a five-objective proposal to one

Aya's proposal bundled five learning objectives: performance anxiety, harmonization, ensemble practice, technical proficiency, and sight-reading. Each was a full product on its own, and shipping all five from an incubator team in one project wasn't realistic.

I scoped the project to performance anxiety, for three reasons that converged. A landscape scan showed existing apps already addressed the technical-skill side of piano, so building another would have been duplicative and a heavy lift for the team. Performance anxiety played to the AVP's strengths (high-fidelity visuals and spatial audio) in a way the other objectives didn't. And it generalized: performance anxiety isn't a piano problem, or even a music problem; it's a near-universal one (presentations, speaking, auditions) which gave the project a much larger scale of potential impact.

In the conversation with Aya, I explained the reasoning and the team's constraints. She was receptive and asked whether the cut objectives could come back later. I didn't promise they would, but I offered avenues: separate engagements down the line, or work she might lead with other partners that we'd support at handoff.

What I learned. Saying no to four objectives goes down easier when the no comes with three reasons converging from different directions (scale, technology fit, team capacity), not one. The conversation isn't "we can't" but "here's why this is the strongest version of your idea."

03

Ran an A/B test instead of letting team intuition decide

Aligning the virtual piano with the real keys was harder than it looked. Our first solution, a virtual piano body with a cutout for the real keys, was clunky in mockup. Manipulating large objects in XR doesn't match how human brains work, and the cutout itself was ambiguous (users couldn't tell how much was meant to show the keys vs. the surrounding piano body).

The team had a strong intuition the approach wasn't working. What we didn't have was evidence for what users would prefer instead. So instead of iterating on the first solution or picking a replacement based on team preference, I prototyped a second approach (a transparent cuboid overlaid on the real keys) and ran an A/B test. The point wasn't to declare a winner internally, it was to let non-XR-expert users tell us what their brains actually wanted, with the team's biases removed from the answer.

Results were mixed enough that neither solution won outright. The team is now designing a hybrid drawing on what each test surfaced.

What I learned. Team intuition is a strong signal that something is wrong, but a weak signal for what's right. When the team agrees a solution is clunky but disagrees on what to replace it with, that's a moment to run the comparison instead of arguing about it.

Outcome

As of May 2026, the working prototype runs on Apple Vision Pro with both venues fully built, live audience, calibrated acoustics, Bluetooth MIDI routing, and the full pre-performance to performance flow. IRB approval is in place. Pilot testing has been conducted. The student guide is deployed and in students' hands. The load-bearing pieces are complete.

Done

✓IRB approval in place (U-M Institute for Social Research)
✓Multiple iteration loops of usability testing completed; findings drove key product decisions
✓Student guide deployed
✓Both venues (Hill + Britton) fully built; live audience and calibrated acoustics running

In progress

○Alignment UX (hybrid solution in design)
○Onboarding sequence and summary screen content
○Instructor-side tooling (planned, not yet built)

Read the deep dive

Methods, artifacts, and full reflections.