A number without conditions is marketing, not measurement.
Half the doctrine is “measured flat”. That is only worth something if the numbers are produced the same way every time and every figure states how it was taken.
Most loudspeaker specifications are written to be unfalsifiable. A frequency range with no tolerance window. A sensitivity figure taken at whichever point on the curve looked best. A distortion number with no output level attached. All of it technically true and none of it useful.
Here is the protocol every model on this site is held to. It is published so you can check the work, and so that anyone who measures a pair independently is measuring the same way.
| Parameter | How it is taken |
|---|---|
| Frequency response | Gated quasi-anechoic, 1 m, on the tweeter axis, gate window stated with the curve |
| Smoothing | 1/12 octave — anything heavier starts hiding problems |
| Low frequency | Nearfield woofer and port summed, spliced at a stated frequency |
| Extension | Quoted at −3 dB and −10 dB, both, never just the flattering one |
| Off-axis | Horizontal and vertical, 0–60° in 15° steps |
| Sensitivity | 2.83 V at 1 m, averaged 300 Hz–3 kHz, never a single peak |
| Impedance | Nominal and minimum both stated, with the frequency of the minimum |
| Distortion | Quoted with the output level and distance it was taken at |
| Pair matching | Both cabinets overlaid, maximum deviation across 300 Hz–10 kHz |
What the target actually is
Flat on-axis, with off-axis output that declines smoothly and gradually rather than collapsing at the crossover. That combination is what decades of controlled listening research keeps landing on, and it is the reason the room you put these in will not fight them.
It is also the target that serves both halves of the audience at once. Flat and well-behaved is what a producer needs in order to trust a decision. It is also, not coincidentally, what stops a speaker becoming tiring three hours into an evening.
What ships with your pair
The measured curve for your serial number, printed on paper, in the box. Not a QR code to a generic PDF of a representative sample — the measurement of the two cabinets you now own, taken after the finish cured, with the deviation between them stated.
Model-by-model figures publish as they are measured. Where a model’s full set is not on the site yet, that is because it has not been measured on the current build — not because it is being withheld. Ask and you will get whatever exists today.
On measurement
Why publish measurements at all?
Because the alternative is asking you to take adjectives on faith while spending four figures on something you cannot audition first. Published curves are the cheapest honest thing a direct-to-customer speaker builder can offer.
What if the measurements aren't perfect?
They won't be. No loudspeaker measures perfectly, and any brand publishing only flawless curves is publishing selectively. You will see the real ones, including the parts a marketing department would smooth away.
Can I get the curve for my specific pair?
You already have it — it is printed and in the box, with your serial number on it. If you lose it, ask; every pair's measurement is kept against its serial.
Ask for the numbers.
Any model, any measurement, whatever exists today. If it hasn’t been taken yet I will say so rather than estimate.