Caliper: Headphone Measurement

Caliper measures headphones on an acoustic fixture and produces a result a stranger can check. Any app can draw a curve. The hard part is knowing whether the same headphone, measured again, would give you the same curve — because if it would not, the differences you are reading belong to how it was sitting, not to the headphone. So Caliper asks for a few captures instead of one, and spends them establishing its own error budget. WHAT IT DOES • Drives a logarithmic sine sweep through one earcup at a time and records both capsules. • Applies your rig's per-capsule calibration, bound by capsule serial, so a calibration from another fixture cannot be used by mistake. • Refuses a capture that fails its own quality gates: signal-to-noise, distortion, clipping, and whether the headphone is actually sealed on the fixture at all. • Selects the medoid of a set of seatings rather than averaging them. An average is a curve no seating produced. • Reports the spread across seatings in four bands, so a curve arrives with its uncertainty rather than without it. • Distortion comes free with the sweep. The harmonics arrive before the linear impulse and cost no second pass. THE DUPLICATE PAIR Two sweeps of one ear with nothing touched in between. Any difference between them is the instrument and the room, not the headphone — which makes it the smallest difference the run is entitled to call real. Every verdict afterwards is measured against it, and a run will not complete without one. SEATINGS Take the headphone off, put it back on, measure again. The spread between seatings is the honest uncertainty of the measurement, and it is usually larger than people expect. How many you need is derived from how much yours actually disagree — an easy headphone may need one, a difficult one eight. PROVENANCE, NOT A BADGE Every measurement records its rig, its coupler baseline, its calibration, its analysis chain and its drive level. The tier — Screening, Repeatable, Reference — is computed from what the file contains and is never asserted. Anyone holding the bundle can recompute every stored number from the archived impulse responses, including ours. The app says what the run still needs and what to do next, right above the measure button. Captures may be taken in any order, and each one counts for whatever it satisfies. RIGS Profiles ship for the miniDSP EARS Pro and the original EARS, GRAS 43AG, Brüel & Kjær 5128, RA0402 and the KB-series. A rig whose thresholds have not been measured on that fixture is marked DECLARED rather than CHECKED, because naming a rig and verifying against one are different claims and the app will not blur them. GETTING YOUR MEASUREMENT OUT • squig.link / CrinGraph text files, so your curve can be compared where people actually compare curves • A measurement bundle that re-verifies every published number from the archived impulse responses • PNG carrying its provenance: rig, fixture and calibration serials, analysis chain, seating count • Hear this in Attune, to listen to the headphone you just measured There is also a simulated fixture for trying the app without hardware. Every capture it makes is stamped, and exporting a run that holds one is refused. FREE, AND NOT A TRIAL Caliper is free with no in-app purchases and no paid tier. A verification tool a skeptic must pay to use is not a verification tool. PRIVACY Caliper requests no network capability at all. Your measurements are files on your own device, and the app has no route to send them anywhere.