ST-01 — the appealing part doesn't work

Case study 04 · Concept · Feasibility · Acoustics · Product scope

The appealing part was the part that doesn’t work.

ST-01 · Acoustic Shot Detection · Rev A · 2026-08-06 · Draft

Can a phone time competition shooting by listening? The answer decides whether the product exists in the shape it was imagined.

Conditional

Not “no,” and the distinction is the whole document. The product survives; the risk moves off the part that cannot carry it.

Why the obvious version fails

155–160dB SPL — a 9 mm muzzle report at one meter

120–130dB — where a phone microphone overloads

Every shot clips. And the splits that have to be resolved run 0.12–0.15 s while an outdoor bay’s reverberation tail runs 50–150 ms — the ringing lasts about as long as the gap it must be told apart from.

Play the start beep and detect it in the recording. Its onset is then zero, and no device or Bluetooth latency has to be reconciled.
Gate each shot against the previous one’s decaying tail rather than a fixed threshold.
Ship microphone detection for practice; read a purpose-built timer over Bluetooth for matches. The hard case is handed to hardware built for it.

The case it cannot settle on paper

A flagged peak that is either a berm reflection, the bay next door, or a live round it cannot afford to discard. Field recordings are being collected at a Utah range by a certified Range Officer to resolve exactly that. The unsolved case is on the figure, not left out of it.

Fig. 1 — Shot string, outdoor bay. Start signal at t0, five confirmed impulses, splits in seconds. The flagged peak is the design problem: berm reflection, adjacent-bay report, or a live round the detector cannot afford to discard.

The last section names the real product — match management, offline-first — names the entrenched free incumbent honestly, and picks a gap beside it rather than a fight with it.

It still says “Draft,” and that is accurate rather than abandoned.

My role — concept, product scope, acoustic research, technical writing.
Status — Rev A, 2026-08-06, Draft. Field recordings being collected at a Utah range by a certified Range Officer.
Written with Claude — authored, not just used. I set the questions and the standard of evidence; the sheet is what came back and what I was willing to sign.

Five sections, five shapes — a verdict, a measurement, a decision list, an unresolved flag, and plain prose. The variation is the point: eight identical rectangles in a column is a form, not a document.

The full data sheet — Rev A, 2026-08-06 →

The same habit, one level down → — the visual proofs this method came from: build one, look at it, then decide.