Who documents their own gear

Every number on this site is supposed to be traceable to something the manufacturer published. Keeping that promise meant recording, for each product, whether the figure existed at all. This page is what fell out of that: a count, per manufacturer, of how often the answer was yes.

It is not a quality ranking. A maker can document everything and build something dull, or document nothing and build something wonderful. It is a measure of one narrow thing — whether you can check a claim before you buy.

What is being counted

One fact per pair of headphones: does the maker print a sensitivity figure with a reference unit attached — per milliwatt, per volt, or at a stated drive voltage? A decibel number with no reference unit is not a specification, because the two readings are about five decibels apart at 300 Ω and eighteen at 16 Ω.

Three facts per source: is there a power figure traceable to something they published, an output impedance, and a noise figure? Each of those decides a different half of whether a source suits a particular pair of headphones.

The count below is strict in two ways. A figure taken from an independent measurement rather than a document counts as not published, however good the measurement. And a maker who publishes power at one load only still counts as having published power — the derivation from there is this site’s problem, not theirs.

Manufacturers with at least six facts to count

ManufacturerIn the databaseFrom their own materialShare
64 Audio10 hp10 / 10100%
Chord3 src9 / 9100%
Etymotic6 hp6 / 6100%
JDS Labs4 src12 / 12100%
Lake People2 src6 / 6100%
Shure15 hp15 / 15100%
SPL3 src9 / 9100%
Symphonium7 hp7 / 7100%
Violectric3 src9 / 9100%
ZMF11 hp11 / 11100%
Audio-Technica27 hp26 / 2796%
FiiO13 hp + 30 src99 / 10396%
Audeze21 hp20 / 2195%
Fostex17 hp16 / 1794%
Focal16 hp15 / 1694%
AKG14 hp13 / 1493%
Dan Clark Audio11 hp10 / 1191%
Truthear8 hp + 1 src10 / 1191%
Sennheiser28 hp25 / 2889%
Moondrop36 hp + 3 src40 / 4589%
Topping6 src16 / 1889%
Austrian Audio8 hp7 / 888%
RME8 src21 / 2488%
Grado15 hp13 / 1587%
Benchmark2 src5 / 683%
Meze17 hp14 / 1782%
Arturia4 src8 / 1267%
Astell&Kern4 src8 / 1267%
Ferrum2 src4 / 667%
IK2 src4 / 667%
Matrix2 src4 / 667%
Roland3 src6 / 967%
Sony12 hp + 4 src16 / 2467%
SSL3 src6 / 967%
Tanchjim6 hp + 3 src10 / 1567%
Tascam2 src4 / 667%
Zoom4 src8 / 1267%
Schiit15 src29 / 4564%
Shanling11 src21 / 3364%
SMSL11 src21 / 3364%
iBasso10 src19 / 3063%
Focusrite7 src13 / 2162%
Audient11 src19 / 3358%
Apogee3 src5 / 956%
Fosi Audio3 src5 / 956%
HiBy5 src8 / 1553%
Aune2 src3 / 650%
Campfire Audio10 hp5 / 1050%
Creative2 src3 / 650%
iFi8 src12 / 2450%
Luxury & Precision2 src3 / 650%
Cayin5 src7 / 1547%
Dunu12 hp + 1 src7 / 1547%
Kiwi Ears12 hp + 2 src8 / 1844%
MOTU4 src5 / 1242%
EPZ4 hp + 2 src4 / 1040%
Beyerdynamic36 hp13 / 3636%
ddHiFi7 src7 / 2133%
HeadAmp2 src2 / 633%
Philips6 hp2 / 633%
PreSonus4 src4 / 1233%
Qudelix2 src2 / 633%
Steinberg4 src4 / 1233%
Tempotec3 src3 / 933%
Universal Audio5 src5 / 1533%
xDuoo4 src4 / 1233%
Simgot7 hp2 / 729%
Questyle3 src2 / 922%
Hidizs8 hp + 5 src5 / 2322%
Geshelli2 src1 / 617%
Gustard2 src1 / 617%
Tangzu7 hp1 / 714%
HiFiMAN28 hp + 2 src2 / 346%
Apple2 src0 / 60%
Little Dot2 src0 / 60%
STAX6 hp0 / 60%
Thieaudio12 hp0 / 120%

Makers with fewer than six countable facts are left out of the ranking, because one product with one missing figure would put them at the bottom of it. They are listed below and have pages of their own.

One caveat the table cannot express: a maker whose entire catalogue is electrostatic — STAX and Warwick Acoustics in this list — scores zero here because electrostatics are specified as a sound pressure at a bias voltage of a hundred volts or more, which is not a sensitivity in the sense this count understands. It is a limit of the question, not a verdict on the maker.

Too few products to rank, one page each

Where these numbers come from

Nothing here is a new measurement or a new claim. Each row is a tally of what is already recorded on the product pages, where the document behind every figure is linked. The reasoning behind the labels is on the method page, and the specification contradictions that turned up along the way are written out on the spec sheet audit. The whole thing downloads as CSV and JSON.

↑