Headphone frequency response comparator

A headphone's curve says more about its sound than any spec sheet, as long as you compare it properly. This headphone frequency response comparison tool overlays curves from our library (approximate Harman targets and two demos) or your own measurements exported from REW, as .frd files or as AutoEq CSVs. It smooths, normalizes and shows the difference between two curves, and against a target it works out how far each band deviates and which parametric EQ would close the gap. Everything is drawn in your browser and your files are never uploaded.

Comparing properly means using curves from the same measurement rig. Across three public measurements of the same open-back model, the Sennheiser HD 600, the curves agree within about 3 dB below 1 kHz and drift up to 15 dB apart above 8 kHz. That's why the tool warns you about it, and why its automatic EQ places no filters above 10 kHz.

How to use the headphone frequency response comparison tool

A comparison takes under a minute and needs just two curves, a reference and the one you care about.

1. Add a curve from the library, for example the approximate Harman over-ear 2018 target, and a second one to compare it with.
2. Upload your measurement as plain text with two columns, Hz and dB. REW's "Export measurement as text", an .frd file or an AutoEq CSV (frequency,raw) all work as they are.
3. Leave smoothing at 1/12 octave and normalization at 500 Hz, so level differences don't hide the shape of the curves.
4. Turn on the difference view to see only what changes between two curves (A − B).
5. Under "Analyze against a target", pick your measurement and the target. You get the deviation per octave band and up to six parametric filters you can copy in Equalizer APO format.

How to read a headphone graph

A frequency response graph shows the level, in dB, a headphone produces at each frequency from 20 Hz to 20 kHz on a logarithmic axis. Once normalized, what matters is the shape, region by region, compared with a target.

The range printed on the box ("5 Hz-40 kHz", "20 Hz-20 kHz") tells you nothing about the sound. Without a tolerance in dB, it only says the headphone reproduces something at those extremes, at any level. The curve tells you how much.

Can you compare graphs from different websites?

For open-back headphones, only below about 1 kHz. We took three public measurements of the Sennheiser HD 600 (by oratory1990, InnerFidelity and Headphone.com, collected in the AutoEq repository), aligned them to their 200-2,000 Hz average and smoothed them at 1/12 octave with this comparator's own code. They are three different units measured years apart, so part of the difference belongs to the units themselves.

The cause is physical. Above roughly 2-3 kHz the wavelength shrinks toward the size of the ear canal and the pinna, so the artificial ear, the coupler and how the headphone sits on it all change the reading. That's why every reviewer keeps their curves in a database of their own, measured on one rig. To compare two models, use two measurements by the same reviewer. To EQ, trust the curve up to about 8-10 kHz and your ears beyond that.

What the Harman target is, and why it isn't flat

The Harman target is the curve most listeners preferred in blind tests. In the over-ear approximation this tool includes, it sits about 5-6 dB above 500 Hz in the bass and about 9 dB above it around 3 kHz, and the in-ear version rises roughly 10 dB in both places. It came out of research by Sean Olive, Todd Welti and their colleagues at Harman, which found that what makes a loudspeaker sound good also applies to headphones. The preferred curve resembles the in-room response of a neutral loudspeaker in a typical listening room, and it was later refined through listening tests, over-ear in 2018 and in-ear in 2019.

The target reflects an average preference. A 2019 AES convention paper by Olive, Welti and Khonsaripour split listeners into three groups, with 64% preferring the target as is, 21% wanting slightly less bass and 15% wanting more. If your taste falls in that last group, a headphone a little above the target in the bass has nothing wrong with it. It's deliberately not flat, because a curve that measured flat at the eardrum would sound dull and lifeless. It would lack the boost the ear canal naturally gives any real sound around 3 kHz.

When it pays to compare headphones by their curves

It pays off most when you choose between two models, build an EQ, swap earpads, measure your own pair or fit a set of in-ears.

Choosing between two headphones before you buy

Download both curves from the same source in the AutoEq repository, which gathers public measurements from oratory1990, Crinacle, RTINGS and many other reviewers, and load them here next to the target. Two headphones measured on the same rig make the fairest comparison you can get without trying them on. Look first at how high the bass shelf sits relative to 500 Hz, how the 2-5 kHz ear gain region compares with the target and whether there are narrow peaks between 5 and 10 kHz, which usually mean sibilance. Two curves within 2 dB of each other up to 5 kHz will sound similar, while a 6 dB difference in the bass is obvious to anyone. If both look good on paper, remember the preference groups in Harman's research, because the "best" curve is the one that matches your taste. Avoid deciding on graphs from two different sites. With open-backs they may agree below 1 kHz and split far apart in the treble, and with closed-backs or in-ears the difference in seal between rigs moves the bass too. If one of the two models is missing from that source, look for another source that has both rather than mixing.

Building a headphone EQ with the comparator's AutoEQ

With your measurement and a target loaded, the analysis panel proposes up to six peaking filters with Q 2.0, between 20 Hz and 10 kHz, and their combined boost never exceeds 6 dB at any frequency. Copy them in Equalizer APO format along with the preamp line the tool works out, which lowers the level just enough for the boosts not to clip. The limits are deliberate. Above 10 kHz the reading depends so much on the rig and on placement that a precise correction would chase errors, and large boosts eat headroom your source may not have. The same logic runs through our guide to EQ settings by genre, which explains why it pays to cut before you boost. The target matters as well. The approximate Harman curve in the library is defined on Harman's own test fixture, so if your measurement comes from a different rig, load a target adapted to that rig (the AutoEq repository collects several) or be wary of the filters above about 2 kHz, where measurements from different rigs start to drift apart. Treat the filters as a starting point, apply them, listen to a few familiar tracks and trim any filter that makes things worse. A good headphone EQ closes most of the gap in the bass and the mids, which is where the measurement is most trustworthy.

Measuring what new earpads change

To find out whether new earpads change the sound, measure your headphones with the old pads and with the new ones and compare the two curves here. Take several readings with each set, reseating the headphones every time, because the fit alone moves the curve, especially in the bass and the treble. Load the average of each set, turn on the difference view and see where they split apart. If the gap is larger than what reseating alone moves, the change is real. Pads usually shift the bass the most, since they change the seal and the volume of air between the driver and the ear, and the treble next, since they change the distance and angle at which the sound arrives. Some manufacturers sell different pad sets precisely to change the sound, and our guide on when to replace your earpads covers a model whose two sets differ by a full octave of bass extension, along with the wear signs and why third-party pads are a gamble. If a reviewer you follow has published the same model with two different pad sets, the difference view shows you the effect without measuring anything.

Measuring your own headphones

To measure headphones you need a coupler that mimics the ear, as well as the microphone. In-ears use an IEC 60318-4 coupler (the "711" type) and over-ears a rig such as the miniDSP EARS, plus a sweep in REW. A phone's microphone won't do, because without a seal and a known acoustic load the curve says more about the microphone than about the headphone. Once you have the text export, upload it here and compare it only with measurements from your own rig, never with a random graph from the internet. Repeat each measurement three times, reseating the headphone, and keep the average, which separates real changes from what the fit alone moves. If you spot a narrow peak and want to confirm it, run a slow tone sweep at low volume, and the peak will stand out as a sudden jump in loudness. Save each export with the date and the conditions (pads, tips, position), because months later it will tell you whether something has changed. We will add our own measurements to the library once our rig is ready, and our page on how we measure at AcusticaX already describes the headphone rig we have planned.

With in-ears, the seal and the tips move the bass

With in-ears, fit matters as much as the driver. RTINGS, which measures every model several times and on several people, notes that a poor seal causes an uneven or drastic drop in bass. A small leak can easily cost 10 dB or more below 100 Hz. Insertion depth also moves the treble peaks, because it changes the length of ear canal left in front of the nozzle. So if your measurement and a published one disagree in the bass, try another tip size before blaming the unit, and compare foam against silicone with the difference view on. Before you EQ a pair of in-ears, fix the fit, or you'll be equalizing a leak. A practical trick is to measure with the stock tip and with one size up. If the bass rises by several dB with the larger one, your ear wants that size. And when you compare against a published measurement, check which tips the reviewer used, since many note it next to the graph.

A promising curve is useless if your source cannot drive them. Before deciding, check in the headphone amplifier calculator whether they will need an amp.

Frequently asked questions