What this is

Our own measurements, made here on a miniDSP EARS Pro: rig, operator, unit and pad state recorded for every curve. Raw, not diffuse-field compensated, and published whatever they say. The corpus is small and growing: we are measuring from scratch rather than inheriting somebody else’s numbers.

The summary below is the short version. The full method is published — the rig, the seating rule, all 19 captures a side, and where we stop trusting our own graphs.

How the headphone is seated

Deliberately, for the best fit, adjusted for seal and for sound, the way you adjust a headphone before you settle in to listen. Not dropped on at random. The common convention averages several careless placements together; we think that answers the wrong question, because you don’t listen at a random seating. What gets published is one real seating, never a blend, chosen by a stated rule.

How repeatable this rig is

Measured in August 2026, and the answer is that it depends on the headphone, not on the rig. The instrument itself contributes 0.002 dB. Lifting the same headphone off and reseating it moves the midrange by 0.06 dB on an HD 800 and 0.32 dB on a PM-3, a fivefold difference between two headphones on one bench. There is no single number to put here, so we are not going to invent one.

Above 2 kHz every headphone we have measured moves more, 0.44 to 0.76 dB, which is why the advice has not changed: treat a line as directional in the treble and don’t read fine detail above about 10 kHz. That used to be a hedge. It is now a measurement.

How much to trust a line

A measurement is one headphone, one unit, on one fixture, seated once. It tells you the shape, not the last half-dB. We have now measured where it stops being trustworthy: below 2 kHz a reseat moves a curve by 0.06 to 0.32 dB depending on the headphone, and from 2 to 10 kHz by 0.44 to 0.76 dB. Read the shape anywhere. Read half-dB detail only in the bass and midrange, and only on a headphone that seats repeatably.

Open headphones read bass-light

That's the fixture's seal leak, not the headphone. It cancels between headphones that seal alike, and doesn't between a sealed one and an open one.

Using these measurements

The measured curves on this page are licensed CC BY-NC 4.0. Quote them, plot them, cite them, teach with them, build on them for anything non-commercial, with attribution to Warren Labs and a link back to this page. Full resolution, on purpose: a measurement you cannot check is not evidence.

The reference shapes drawn for comparison are not our measurements and are not covered. Commercial use, including bundling the data into a product or a service, needs a separate licence from us. Email jamey@warrenlabs.com and say what you have in mind.