What

A hoop on a closet door, and a morning routine: shoot until three makes. Tap an Apple Shortcut, start recording, and call each attempt out loud — “brick” or “splash”. Audio processing on the recording turns that into labelled attempts, consolidated stats, and an automatic daily report.

Why it mattered

It is the smallest honest test of the thesis behind all the independent work: capture should be a byproduct of something enjoyable, not a chore layered on top. Nobody logs free throws in an app for a month. Saying one word while shooting costs nothing and produces a cleanly labelled stream.

It also rhymes with the Imperial dissertation, deliberately. That project instrumented a real-world motor skill with motion capture, VR, and EEG to study how people learn; this one instruments a real-world motor skill with a phone microphone and a shortcut. Same question — how does performance on a real task change over time — at two very different costs. And the capture layer is designed to take more: joint kinematics, a room camera tracking the ball, even neural data recorded during the routine, without changing how the routine feels.

My role

Built all of it: the capture flow, the audio processing, the stats consolidation, and the daily report.

Outcome

Shipped, and in daily use.