back to Audio Technica
back to measurements

published: Jan-19-2021

post separation

NO SMOOTHING is applied to the shown plots. Most measurement sites have some smoothing applied which ‘irons flat’ sharp peaks and ‘wiggles’. I do not use smoothing because some info about sound quality is lost when plots are smoothed.

Aside from a small correction of the microphone itself also some correction in the lowest frequencies is applied to the plots to compensate for the perceived loss of bass when using headphones. This is described HERE in more detail.
A ‘horizontal‘ frequency response curve on the shown frequency response plots on this website thus indicates a perceived ‘flat’ tonal signature.

ALL measurements are made with a good SEAL on a flatbed measurement rig.
The shape of your head, bone structure, pad size, pad ‘softness,  (compliance), hair or no hair and or wearing glasses may (drastically) change the frequency response of some headphones, so… your personal experience may differ substantially from these plots.

Frequency response (tonal balance) is the most sound-determining aspect of headphones. A horizontal line shows audible neutral response in the plots on this website. Deviations in different severities at different frequency bands have an effect on the sound character.
The bigger the deviation the stronger the effect.

Below an aid to help determining the sound character of headphones with relation to the frequency response.


post separation
audio-technica ATH-SR9

The audio-technica ATH-SR9 is a closed headphone launched in June 2017 and was discontinued in 2020. It sold for around € 500.- in the E.U. and $ 450.- in the US.  Between 2019 and 2020 the headphone dropped in price to around € 280.-.  To me still a much too high price for this headphone.

The measured headphone is used and was sent in for measurements as the owner didn’t like the sound.
It has a professional quality look. When handling it, however, everything creeks. There is enough swivel and headband extension so that it will fit many heads. The headband padding and ear-pads feel soft and looks like sturdy leather. It has a synthetic/silicone feel to it. Maybe it won’t flake as fast as other pleather.

Left & Right markings are only found inside the pads. The left & Right connectors can be told apart by a blue ‘ring’ near the connector itself on the left side. You need to the left connector in the left cup. Easy to swap these by accident.
The connector is a slightly changed MCX connector. An MCX connector, however has a (male) pin in the centre, this connector has a female inner contact. So no ‘standard’ available plugs can be used.

Room for the ears is a bit limited. The width inside the pads (35mm) is a bit on the small side. Height and depth are ‘normal’ sized.
Isolation from outside noises is decent. The cups are very ‘microphonic’ so touching the cups while adjusting is quite audible.

The 1.2m cable is thin and ‘springy’. It has a silicone feel to it. An angled 3.5mm TRRS connector allows microphone and remote control. The cable itself, just like the cups are microphonic. At least the first 50cm or so.
The ATH-SR9 was aimed at the portable market. The headphone is quite sensitive and can be driven to loud levels even from a phone. The 1.2m cable with a mic/remote also is a clear indicator it is meant for the portable market.


Type: Over ear, closed
Usage: Home, studio, portable.
Driver type: dynamic
Driver size: 40 mm (specified as 45mm)
Pads: replaceable, pleather (fake leather) with fast memory foam.
Inner pad dimensions: height: 60mm, width: 35 mm, depth: 19mm.
Foldable: No, but can lay flat.
Headphone connector: A2DC connectors (a modified MCX connector)
Cable entry: dual sided
Cable: 3.0m and 1.2m with mic./remote
Power rating: 1.5W (1500mW)
Max. Voltage: 8.2V
Max current: 180mA
Max. S.P.L.  129 dB
Impedance: 47 Ω
Efficiency: 97dB @ 1mW
Sensitivity: 110dB @ 1V
Weight: 270 g.  (without cable, the 1.2m cable weighs 20 gram)
Clamping force: medium-high
Colour: Silver with black pleather headband.
Accessories: Detachable 3m, Detachable 1.2m with microphone and controls, 6.3mm adapter, Hard-case

Sound description:

The overall sound signature of the ATH-SR9 is complex. It has weird sounding mids and is a bit ‘light’ in the bass. The mids are ‘sharp/bright’ and there is a very coarse/grainy and piercing treble up top. Popular recordings sound nasty. It is hard to believe this headphone costs € 500.- (in the US: $450.- ).


Below the frequency response of the ATH-SR9 (Left, Right)

For € 500.- this is an utterly disappointing frequency response. The rising bass levels isn’t the biggest issue. The steep 13dB drop at 200Hz is a bit much. In fact it looks a lot like the HD820 (shown below)

Freq response

When looking at both plots one may think these look very similar so they will sound alike. Nothing could be further from the truth.
Below the 2 plots overlaid which shows quite a few differences.
The HD820 does have a weird bass response but mids and treble sound fine. Not shrill, bright and nasty like the SR9.
The extra ‘presence’ (and this is even measured without a pinna !) from 1kHz to 3kHz and the massive peak (+10dB) at 7kHz followed by broad peak around 15kHz and the sharp dips here and there make the treble bright shrill and grainy.
The guys/gals that approved this headphone must have serious hearing damage if they did not hear this.
Is the headphone defective ? Both channels ? Is this a fake ? doesn’t look like it.

Below the ATH-SR9  compared to the much cheaper and generally well received ATH-MSR7.
The SR9 has a lot more upper bass. Lower bass, however is already 10dB down so lacks ‘body’. The upper mids (above 2kHz) slope downwards on the MSR7 but the SR9 is about 9dB higher at 3kHz where shrillness resides.
Both headphones have a treble peak. The one from the SR9 not only is higher but also lower in frequency adding to sibilance and shrillness. The peaks above 10kHz give a nasty sharp ‘etch’ to instruments with lots of harmonics.


Seal can be an issue with closed-back headphones but is usually less of a problem for open headphones. Breaking the seal (improper fit on the head) usually means a loss of (sub)bass. Below the effect of different levels of seal breach are shown .
Perfect seal, Seal broken with a pair of glasses, seal broken using a 6.3mm TRS plug.Due to the soft pads with their large surface people wearing glasses that sit close to the skin do not have much to fear.
A more substantial seal breach does lower the overall bass levels and make the bass drop-off much sooner.

output resistance / damping-factor

As this is a dynamic headphone the frequency response might be amplifier output resistance dependent when certain higher output resistance amplifiers are used.
To test this the headphone is measured via a low impedance amplifier of 0.2Ω, at 10Ω, 32Ω 120Ω. These traces are overlaid so tonal balance changes can easily be seen. On a higher output resistance amplifier the output level will be lower as well. At 10Ω the level drops 2dB, at 32Ω 5dB and at 120Ω the level is 11dB lower.
The good news here is that a higher output resistance only results in a lower output level. The impedance is as good as ruler flat and the tonal balance is not affected by the output resistance at all. The resonance frequency of the driver is around 100Hz and the level change with a 120Ω output resistance is around 1dB max.

Below the distortion measurements of the ATH-SR9 (Left channel).
The plot above is in a dB scale, below the same measurement but in a percentage scale.The distortion in the lower frequencies is as can be expected from a 40mm driver at 90dB SPL. From 300Hz upwards the distortion is quite low. The resonance at 7kHz isn’t helped any by the distortion reaching an audible 0.5%.
NOTE: The actual 2nd harmonic distortion above 1kHz may well be lower than 0.2%. A shortcoming (measurement limit) of my measurement rig.


Linearity is checked by measuring the frequency response 4x at levels of 70dB, 80dB, 90dB and 100dB. Because of the wonky response and the particular interest in the lowest frequencies, the range where non linearity shows itself, the SR9 got some global EQ to make the response more ‘even’. Then the 4 traces are overlaid so they match at 1kHz. When the driver has issues this causes the FR to roll-off in the lowest frequencies (traces no longer overlapping) at the highest SPL level(s).Clearly there is no evidence of non linear behaviour.


Phase measurements

I usually don’t post these but more data is always welcome. below the phase response of the SR9.
The dip around 350Hz is also visible here as well as the dip at 4kHz and 7KHz.

Below the Group Delay plot some readers like to see.

Some weird behaviour around the lower mids. This clearly isn’t just a frequency related dip but is a resonance issue.

Time domain measurements

Below the CSD of the ATH-SR9 (Left and Right are overlaid)We see some resonance around 2.5kHz, A nasty one at 7kHz sticks out. Note that the dip at 4kHz just seems to have a short delay.
This is also observed in the Group Delay plot. The part above 10kHz also is made up from some resonances.

Another form of looking in the time domain is the spectrum plot. The amplitude is color coded and both the time scale and frequency scale differ as well.

Below the spectrum plot of the ATH-SR9 (Left channel) There is some resonance at around 500Hz and a longer one just above 2kHz. This is the ‘brightness’ area.

Below the step response with a dB scale (so not similar to an oscilloscope plot which has a linear scale).  The step response (Left and Right channel overlaid)

The rising edge reaches 0dB but also rings a lot and then drops. Only to rise again and peaking in the lower mids, upper bass.

The measurements above are indicative of less than optimal sound. Measuring a device like this and just mentioning it is crap is one thing. Let’s see if it can be improved upon, preferably with some simple modifications.


Pads: I have tried a bunch of pads, after-market and original ones from various other headphones that (kind of) fitted.
None of them improved the response impressively enough to warrant a change. Below some of the pads that were used.Original pads compared to SRH840, SRH940, Alcantara, HM5 leather pads. None of these gave improvements or worthwhile changes. In fact the overall sound signature did not change. Clearly the wonky response is not caused by the pads but is a driver/enclosure thing.

Another easy experiment is to seal the baffle ports (the white paper around the driver). Below various degrees of ‘taping off’ parts of the paper covered ports and their effect.All ports open (stock), Only 2 bottom ‘sections’ taped shut, 4 bottom sections taped shut. side and bottom sections taped shut, all sections taped shut. What can be learned from this is that the ports have the most influence in the 1kHz to 4kHz section and the overall amount of bass. The most interesting was the green line where only 2 bottom sections were taped shut. On the left a picture of the 2 taped-shut bottom sections.

This levelled out the mids between 500Hz and 4kHz making the sound less shrill.
The dip and roll-off remained the same.

Another way is to add damping behind the driver. On the right side we see the rear of the driver is already substantially damped by a felt ring covering the holes behind the membrane. Applying more damping, for instance by restricting airflow of the magnet hole will only reduce the lowest parts of the frequency range which is not desired.
The cup itself was just an empty cavity. Just removing reflections from the rear of the cup did not change the measured response significantly enough to make a difference.

As a last resort some damping materials were applied behind the driver in the cup.
The most effective one was a dab of medical cotton.
The material was fluffed-up a bit and just enough was put in the cups so that the entire driver was covered a bit. A picture is shown on the left. My simple scale could not measure the weight of the cotton (not wool or fiber). Replicating this means one may have to experiment a bit. A bit too dense means bass levels were lowered too much. Too little gave less effect. It is important to have the same amount in both cups.

The effect of this cotton is shown in the plot below. Stock, but using SRH-940 pads because these pads are really easy to take off and put on and did not change the frequency response too much versus the same situation but with cotton in the cups.

This gets the response quite a bit flatter without sacrificing bass extension. the way too high bass levels are lowered almost 4dB and above all the dip around 300Hz became a lot less deep. Also around 3kHz a similar effect like taping off 2 sections of the ‘ports’ is seen. This is logical as the sound from the rear of the driver is also less effective now.

The treble peaks are hardly affected. This should be done on the front side of the driver. Some 2mm thick wool-felt (from a crafts tore) proved to e the most effective.
A piece of felt was cut that can simply be put in the recessed ‘chamber’ for the pads. There is no need to secure/fix this as it nicely fills the space between the dust netting in the pads and the driver. Just put it in and mount the pads and it stays where it needs to stay.

Below the effect of the 3 (easily reversible) modifications combined. That is the 2 taped-shut sections, the cotton in the cups and the wool-felt in front of the driver using the original pads (Left, Right).That is quite an improvement. Bass response, while still a bit elevated  between 100Hz and 200Hz is better balances and is much less downwards sloping towards the subbass. The dip around 300Hz is substantially smaller. The area between 600Hz and 4kHz is much more even and not sloping upwards so shrillness is removed. The felt lowers the 10-20kHz peak to desired levels. While the 7kHz peak is greatly reduced already it still is audible. The overall sound, however, is already greatly improved.

The easiest way to address the remaining treble peak is by using a passive filter. The circuit diagram for this filter is shown below.

Below a filter section mounted in the cup. It is fixed onto the driver. The – of the driver is connected via the filter to the white wire and then soldered to the existing wire and then heat-shrink tubing is used to isolate the solder-joint. The red wire was connected directly to the driver.

The effect of the filter is shown in the plot below. Modified but without the filter and modified but with filter.When all these modifications and filter is applied we get the response as shown below.The frequency response is much improved. The treble quality, while greatly improved, remains somewhat ‘coarse’ and doesn’t become refined. The overall change between stock and modified SR9 is shown below.
The response from the modified SR9 looks quite similar to that of the MSR7. Note: the SR9 response is shown without the treble filter.

Below the CSD of the modified SR9 + treble filter.

and below the step response of the modified and filtered SR9.


The audio-technica ATH-SR9 is a very expensive and poor sounding headphone. I cannot imagine who at audio-technica thought this was a good headphone worth that kind of money.
The sound is bass shy and ‘different’ from other headphones. It sound shrill, coarse/grainy and sharp.

With the described modifications the sound can be changed for the better for a minimal amount of materials (some woolfelt, cotton and tape) This makes the headphone more normal sounding though still not a high flyer.
Those owning this headphone and not able to return them or don’t want to give it away or trash it have an alternative now.

post separation

back to Audio Technica
back to measurements


  1. Pawel Liedtka says:

    Thanks Solderdude.
    Wow, this is bad. I was expecting something really decent from AT at this price point.
    The cotton damping makes a huge improvement. Better to get the M50X, even though the pads have to be changed every 2-3yrs or so with heavy use. Again, I would try some cotton inside the cups to flatten the freq response.

  2. Solderdude says:

    Yes, the SR9 is quite the disappointment. With the mods it wasn’t that terrible any more.

  3. Oliver Leang says:

    Had I read this sooner I would’ve convinced myself not to get them. Thanks a lot for sharing the mods.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.