What Headphones do not need (or scream for) EQ?

[ADU]

Some diffuse field compensated graphs of the AKG K371, based on the Sound Guys raw HBK 5128 measurements here...

AKG K371 review

The AKG K371 headset offers stellar audio performance and is uniquely comfortable. It may not be perfect but, boy, is it close.

www.soundguys.com

First graph compares the AKG K371 in black to the average sound power of 10 well-extended neutral loudspeakers in orange.

Next graph compares the K371 to the average sound power of the 5 best extended loudspeakers in my current sampling (which will be further expanded soon, I hope!), which is shown in magenta.

The horizontal gridlines on the graphs represent 1 dB steps. And both are normalized in the upper bass to hopefully give a little better idea where the K371 is conforming or tracking well to the SP curve, and where it's not. The average of the 5 best extended speakers (in magenta) has a slightly steeper overall slope than my previous sound power curve, because of a little better extension and elevation in the sub-bass, and a little more rolloff in the high treble (possibly due to the small sample size?).

The K371 is a headphone that I have listened to quite a few times. And my overall impression is that it was fairly neutral, but it also lacked some warmth in the bass, and air in the higher frequencies. And it had a bit of glare or harshness in the midrange or low treble (I wasn't quite sure which when listening). And it's sub-bass seemed a little overdone.

If you assume that sound power is a reasonably good target for a neutral headphone's DF response, then it isn't too difficult to see why the K371 might have given me some of these impressions, based on the above comparisons.

The K371 is tracking the SP curves well in some places. And less well in others. The ear canal resonance at ~15 kHz is substantially reduced (or inverted?) on the K371. And so the upper treble rolls off too quickly. This may be consistent with the Harman target (which is also too rolled off in the same area imho). But it is not consistent with the response of some other high quality headphones. And frankly just sounds wrong to my ears, because there's not enough detail up top.

To be fair, I am beginning to get up in years, and my high frequency hearing is undoubtedly a bit impaired at this point... I think Dr. Olive has more or less conceded though that the Harman target may not be as reliable in this range. And so may underestimate a neutral response in the HF a bit. (This is the impression I've gotten anyway.)

The K371 appears to be tracking the SP curves fairly well in the mid and lower treble though, except for a small notch at around 6 kHz where some notes may get lost. It frankly surprised me a little, to see just how well it's doing in this area. Because I thought some of the glare I heard might be coming from this range as well.

I think the impression of glare might be the result of a combination of factors though, including the lack of balance/air in the upper treble, the more forward mids, and also the somewhat depressed level of the bass versus the mids. The depressed bass and higher frequencies means that the FR on this headphone is somewhat biased towards the upper midrange (especially around 1.5 kHz, it seems), and to a lesser extent the lower treble... Which seems to be a recipe for glare.

The K371 does appear to have a fairly strong/solid peak in the mid-treble at around 10 kHz though, in between the ear canal resonances that would normally be situated around 8 and 15 kHz. So there could be just a hair too much energy in that 10k spot. If there is a bright spot there, it's not glaringly obvious on the plots... so I need to see the responses of more headphones to better assess what's goin on there.

The lack of apparent warmth in the bass seems to come from two things. The dip at around 80-90 Hz. And the generally depressed level of the bass, versus the more forward sub-bass and mids.

It may be nice to have such good extension in the sub-bass for some applications. I think a higher resonant frequency in the bass would deliver better overall tonal balance though. The K371 just sounds a bit too limp to my ears in the mid and upper bass, which gives it a cooler sound than I normally like. Some other folks seem to like the fact that there's no bleed from the bass into the mids though on this headphone.

The AKG K371 is not what I'd call a really bad sounding headphone without EQ. I just wish it had a little more air, and more warmth/body in the bass, and a little less glare in the upper mids/low treble.

 

[ADU]

Some similar DF compensated plots of the AudioTechnica M50x. These are also based on Sound Guys HBK 5128 measurements. And the orange and magenta sound power curves are the same as in the previous K371 plots above.

M50X compared to a -1.25 dB per octave slope...

 

[isostasy]

@ADU I am quite confused about your methodology here. I kind of get why you're showing diffuse field compensated targets, because this sort of shows the downwards slope of an in-room loudspeaker target, giving you some indication of what the headphone will sound like, but diffuse field is simply not what these headphones are tuned for. The whole point of using the Harman target is you don't need to mess around with diffuse field compensations, guessing at the steepness of the slope you want compared to different scenarios. It's then much easier to compare headphones because you don't need to mentally visualize some kind of slope, you can just compensate it and look at where it diverges form the straight line (this is what Amir does in his reviews in his 'Deviation from Preference Graph'). Added to this you're using 5128 measurements with their funny treble response which I can't ever see being replicated in your sound power lines.

I've read some of the previous discussion on this thread and I'm left slightly baffled. Even if you're not technically wrong (which I'm not sure of), I'm certainly very confused by all these various diffuse field compensations and loudspeaker targets you've obfuscated the measurements of headphones with, when we have a very well researched and validated headphone target to use.

 

[ADU]

I understand, and can appreciate some of your confusion on this, isostacy.

I was confused about diffuse field measurements for quite awhile as well. And I also agree with you that the treble on a DF compensated headphone FR curve will probably never precisely match or track a diffuse sound power loudspeaker curve... just like the treble on a raw headphone FR curve will never precisely match the Harman target in that range.

Since both types of targets are based on fairly coarse approximations in the higher frequencies, they'll never capture all of the detail that we'd really need or like to have to more properly and precisely analyze a headphone's (or other in-ear stimulus's) response in that range. We'll probably have to wait for more in-ear measurements of actual loudspeakers in semi-reflective rooms to really begin to better understand what should probably be happening there.

I think that using DF compensation with a target based on sound power has the potential to get a few steps closer to this objective though. And I'm basing that opinion on hundreds of headphone and now also loudspeaker measurements that I've examined over the years, from a variety of different measurement rigs and platforms. (Something which began a number of years go with Tyll Hertsen's great HeadRoom graphing tool!)

Another thing that I like about an approach that uses something like diffuse field compensation is that it has a greater potential for platform independence. Because it's based on a common standard or stimulus used by many different HATS measurement systems for the purposes of measurement calibration.

The standard for a neutral in-ear response needs to be more portable between different measurements systems/platforms. And compensation or calibration based on a common stimulus (such as a diffuse sound field) is one possible way of making that a bit easier. This is an idea which unfortunately fell by the wayside in the wake of the Harman research that is something I think we need to get back to.

I'd like to see some even better standards for this though. Including some new standards based on the measured in-ear response of a typical stereo speaker arrangement and listening space in a home... Until that happens though, I think a diffuse sound field measured from inside the ear, and modified with a sound power curve is probably the next best thing. (That's what most of the measurements seem to be telling me anyway.)

Harman's over-ear headphone target actually helps to demonstrate this. Because if you take the raw Harman target, and apply an appropriate diffuse field correction to it, it will closely approximate the diffuse sound power response of a well-extended neutral loudspeaker.

I've previously posted examples of several different loudspeakers which help to illustrate this here. And can post some more, if anyone would like to see them...

Master Complaint Thread About Headphone Measurements

Together we can rule the headphone measurement world ! Pass, but thanks.

www.audiosciencereview.com

I think the four examples (which include the Dutch & Dutch 8C, Infinity Prelude MTS, Mesanovic RTM10, and Genelec 1032A) in the link above probably demonstrate the similarities sufficiently though.

These are speakers that are generally well-regarded for their flat on-axis response, neutrality, and also their extension into the lower frequencies. So they were good examples to demonstrate this imo. I see that the original links to the spinorama graphs for these speakers are broken now though. So here are some updated links for them...

A collection of loudspeakers measurements

pierreaubert.github.io

A collection of loudspeakers measurements

pierreaubert.github.io

A collection of loudspeakers measurements

pierreaubert.github.io

A collection of loudspeakers measurements

pierreaubert.github.io

 

[isostasy]

@ADU hmm I think I'll opt for agreeing to disagree so as not to derail the thread.

My general answer to the question would be to agree with @solderdude , that most headphones you put on your head won't scream for EQ immediately and can be enjoyed without it. I'm reminded of Tyll's DIY headphone competition: 2nd place for 'worst measuring', he writes 'After seeing these measurements I thought for sure they would be unlistenable. Oddly enough, they sounded far better than they measured.' We don't give our brains enough credit for adjusting to what comes in through our ears.

From personal experience:
- I find Etymotic ER2XR to be perfect;
- Sennheiser HD6XX can sound better with a sub-bass boost/mid-bass cut but would not call this essential;
- Grado SR80i with taped L cushions and craft felt before the driver sounds great to me, there is still a small peak at 2kHz but it's oddly only noticeable when you cut it out;
- Sennheiser HD451 is a weird one which is enjoyable despite obvious deficiencies in frequency response.

 

[ADU]

@ADU hmm I think I'll opt for agreeing to disagree so as not to derail the thread.

Understood. I probably shouldn't have to tried to explain all of the above again anyway. Because it probably just bores some people here.

DF+SP is, in a nutshell, just another method of replicating or approximating a neutral response curve (like the Harman target) for other measurement systems than the GRAS. In some ways, it is similar to the process Jaakko used for his AutoEQ project. I think DF+SP has the potential to be a bit more accurate, convenient, reliable, flexible, and also portable though. At least with some of the newer HATS systems, like the HBK 5128, where there's good diffuse field data readily available.

I hope that either the users, or the manufacturers and distributors of the newer HATS rigs will try to do some in-ear measurements of actual loudspeakers with their rigs though, in semi-reflective spaces like in a home. So we can see how that actually compares with the Harman target, and with the raw in-ear measurements of good headphones, and also my DF+SP approach. Because I think that would give us a much better idea of what to look for in the way of a neutral in-ear response on their setups. And it would also add alot of value to their systems, and to the reviews and measurements made from them by the headphone community.

 

[solderdude]

What audioholics says about sound power:

the sound power curve isn’t a great guide of sound quality...

Sound power is not a good measure as it is a measurement of total sound pressure coming from the speakers without special regard for the sounds directed at you and also assumes full reflectivity of the room in all directions.
When you put a microphone on a listening spot you measure both direct and reflected sound. As long as that is in phase (low frequencies) it adds to the total sound but as the frequencies get higher the brain 'knows' what is reflected and direct sound and perceives it differently as the sound power would suggest.
Also the sound power will be very room dependent.
Toe-in and placement as well as the room and all that is in it will make the sound power plot be quite different from the actual total power per frequency for each speaker and room. For this reason alone it is not a useful starting point.
If it were all speaker and headphone producers would use this as a target, yet none are. It is measured but as a tool amongst measurements.

On top of that diffuse field is either measured in a fully echoic room or simulated in an anechoic room with speakers placed all around that all emit the same sound.
This is completely not what headphones do. It was used in headphone measurements with HATS because it was closer to how the headphone was perceived as that from free-field (in front or at an angle) in an anechoic room. Those were the only 2 available 'compensation curves'.
Because of the somewhat poor relation between DF and perceived sound better matching methods were needed. Harman's attempt is one of them which also includes room effects (of somewhat 'normal' room effects) and average preference of listeners.

So... no I don't think any headphone manufacturers nor measurements will ever be based on sound power and DF measurements. I also don't think it is a good idea as sound power is not accurately showing/representing how a speaker is perceived in a normal room only how that speaker measures in an echoic room (added power of sound).
You are just substituting the 'elevated lows' from the sound power measurements combined with DF measurements for the 'elevated bass from preference' and 'room alike treble reduction of speakers in a room'.
Because of this (both Harman and your proposed method) will result in elevated bass and slight subdued treble but not following the same line as they are based on different methods.

I know this is how you see it and think there is a high correlation but there is a much better correlation with Harman or optimum hifi (when you are not into Harman boost)

There is even a thread for it and this discussion doesn't really belong in a thread about headphones that can be used and sound good enough (to whom ?) without any EQ.
Perhaps post your ideas in one of the many 'headphone measurement technique threads'.

 

[ADU]

Thank you for chiming in on this, solderdude.

Fwiw, I understand many of the objections that you're raising here. And I hope maybe someday soon that I'll be to explain this in a way that you and some other engineers can better understand, and accept.

The DF+SP approach works very well for me though, as a shorthand method for determining a measurement system's neutral response. So I'll continue to use it. And also try to explain to those who are interested why I think it works so well, based on the measurements, when they ask about it, or question it, as some have done again here.

What audioholics says about sound power:

Sound power is not a good measure as it is a measurement of total sound pressure coming from the speakers without special regard for the sounds directed at you and also assumes full reflectivity of the room in all directions.
When you put a microphone on a listening spot you measure both direct and reflected sound. As long as that is in phase (low frequencies) it adds to the total sound but as the frequencies get higher the brain 'knows' what is reflected and direct sound and perceives it differently as the sound power would suggest.
Also the sound power will be very room dependent.
Toe-in and placement as well as the room and all that is in it will make the sound power plot be quite different from the actual total power per frequency for each speaker and room. For this reason alone it is not a useful starting point.
If it were all speaker and headphone producers would use this as a target, yet none are. It is measured but as a tool amongst measurements.

This has already been addressed in the Master Complaint Thread. And I don't really have anything new to add on it at this time.

On top of that diffuse field is either measured in a fully echoic room or simulated in an anechoic room with speakers placed all around that all emit the same sound.
This is completely not what headphones do. It was used in headphone measurements with HATS because it was closer to how the headphone was perceived as that from free-field (in front or at an angle) in an anechoic room.

Yes, it was better than free field. And it was (and still is imo) a useful compensation tool to remove the HRTF of a measurement rig, and to make measurements more cross-platform compatible.

Those were the only 2 available 'compensation curves'.
Because of the somewhat poor relation between DF and perceived sound better matching methods were needed. Harman's attempt is one of them which also includes room effects (of somewhat 'normal' room effects) and average preference of listeners.

I agree that better matching methods were needed. And for those who are unable (or unwilling) to use the Harman target, using DF with sound power can be a good alternative. And the measurements bear this out (imo).

So... no I don't think any headphone manufacturers nor measurements will ever be based on sound power and DF measurements.

Needless to say, I disagree with this. Because I'm already using it for measurements. And many headphone manufacturers are already designing headphones to meet this spec, whether they know it or not... based on the measurements.

I know this is how you see it and think there is a high correlation but there is a much better correlation with Harman or optimum hifi (when you are not into Harman boost)

Yes, I do. And I have no objections to people using Harman or other targets, such as Oratory1990's Optimum HiFi, if that's the type or response they want.

I think the correlation will be better with the DF+SP model though, because it's flexible enough to include a variety of different bass targets, depending on the type of speaker and sound power response you're trying to model.

There is even a thread for it and this discussion doesn't really belong in a thread about headphones that can be used and sound good enough (to whom ?) without any EQ.

Maybe it does and maybe it doesn't. Someone raised some questions about the approach I was using to determine whether or not EQ was necessary though, and so I answered them. So I guess that sort of makes it relevant.

 

[Thomas_A]

The Beyerdynamic DT150 with DT100 pads, HD600/650, BOSE QC25/35 sounds fine without EQ. DT770 Pro to much energy in treble. As always there are individual differences.

Harman curve without ear canal? HD650, DT150_100P, DT770 Pro 250 Ohm, Bose QC25 measured

The Harman curve has alot of smoothing in the treble. And neither the Rtings nor the Harman targets are that accurate or reliable above a certain range in the treble and upper mids imho, beyond some very broad strokes. So I wouldn't put much faith in either of them, especially in that range...

www.audiosciencereview.com