What measurement tells you about the level of detail a headphone can provide?

[hans132]

I think this site is great and I appreciate the objective approach amir takes to review audio gear without the use of nebulous language. However I have owned and listened to many headphones and checking out the reviews here, I have problems relating my experience to the measurements.

Specifically, what measurement (if any) will tell you something about the level of detail or transparency a headphone can provide?

As an example, I own both the recently reviewed ATH-M50X (well I have the non-x version) and the DT990Pro. In the reviews the ATH-M50X shows better latency as well as better distortion measurements, but as much as I like the ATH-M50, in terms of transparency I think it is at least a league below the DT990Pro. The Beyerdynamic is far better in that regard.

How would I get that information from reading the measurements?

 

[hans132]

Sorry for pushing the thread. Did I post this by any chance in the wrong subforum? Or is there a different reason why I didn't get a response? Should I try the General Audi Discussion or Psychoacoustics subforum? I am still new to this forum, so please bear with me.

 

[Robin L]

As regards headphone measurements. Your ear, your ear canal, is analogous to the room a speaker is in. The room will determine what sort of speaker would be ideal, it has that much impact on the sound. In the same way, it would be difficult to determine the best sounding headphone for you based on measurements. You really have to slap them on your head and listen to them to find out what works for you.

There's a few headphones that Amir has reviewed that more or less measure well, I own two. The Drop 6XX is the same headphone as the Sennheiser HD 650, has low distortion and conforms to the Harman curve until the bass goes below 100hz. The [cheaper and easier to drive] AKG K371 more closely conforms to the Harman curve, has more bass without EQ. Some have complained about build and comfort with the AKG 'phones, but they're my favorites now. Both headphones have relatively low distortion. You might want to look at Amir's measurements for headphones here or at Solderdude's measurements at DIY Audio Heaven. I've heard the AT headphones, ok but did not fall in love with them. On the other hand, never heard Beyer headphones I liked. Different strokes . . .

 

[hans132]

Please don't get this the wrong way, but I feel like you didn't really address my question, so let me try to differentiate:

I am not asking what a good/bad headphone would be. I am explicitly not talking about frequency curves and listening preferences. I own enough headphones to know what I like and am not looking for a new headphone in particular. I am not looking for buying advice. So please tell me if this is the wrong sub-forum to ask my question.

Instead I would like to know how to relate the measurements in amir's and other's reviews to a single metric. How to relate it to the perceived level of detail or transparency of a headphone (I am not sure if there may be a better term). So in the same way the distortion wouldn't matter, if I ask about max volume, it seems to me that the frequency response doesn't matter when I talk about transparency.

From your response I am not sure if you suggesting, that the ear canal affects the perceived detail. If you do, then I fail to see how that would be relevant? A less transparent speaker in a terrible room, would still sound less transparent than a more transparent speaker in the same room.

I was assuming that latency and distortion would affect the transparency (which they might), but seeing, that the notably less transparent headphone (ATH-M50) measures better in both categories than the far more transparent headphone (DT990Pro), there has to be some other measurement or aspect.

From what I read on the forum, there seem to be a lot of people here much more knowledgeable than me when it comes to audio science. My observation seems very basic to me, so I am still hoping that someone can help me make sense of the situation. So I pose my question once again:

How can a headphone sound less transparent than another one, despite measuring better in terms of distortion and latency?

 

[Robin L]

Please don't get this the wrong way, but I feel like you didn't really address my question, so let me try to differentiate:

I am not asking what a good/bad headphone would be. I am explicitly not talking about frequency curves and listening preferences. I own enough headphones to know what I like and am not looking for a new headphone in particular. I am not looking for buying advice.

Instead I would like to know how to relate the measurements in amir's and other's reviews to a single metric. How to relate it to the perceived level of detail or transparency of a headphone (I am not sure if there may be a better term). So in the same way the distortion wouldn't matter, if I ask about max volume, it seems to me that the frequency response doesn't matter when I talk about transparency.

From your response I am not sure if you suggesting, that the ear canal affects the perceived detail. If you do, then I fail to see how that would be relevant? A less transparent speaker in a terrible room, would still sound less transparent than a more transparent speaker in the same room.

I was assuming that latency and distortion would affect the transparency (which they might), but seeing, that the notably less transparent headphone (ATH-M50) measures better in both categories than the far more transparent headphone (DT990Pro), there has to be some other measurement or aspect.

From what I read on the forum, there seem to be a lot of people here much more knowledgeable than me when it comes to audio science. My observation seems very basic to me, so I am still hoping that someone can help me make sense of the situation. So I pose my question once again:

How can a headphone sound less transparent than another one, despite measuring better in terms of distortion and latency?

What I said: sound is really more about your brain anyway. With headphones, that impression is also going to depend on one's ears as well. What is "more transparent" to one person is "shrill" to another, "dull" to a third. So there won't be a single metric for "Transparency". Frequency response is a measurable metric, distortion is a measurable metric. "Transparency" is a subjective impression, not something measurable.

 

[Jimbob54]

Couple of points I want to clarify with you first (and I am a million miles away from technically competent in this regard.

1. What do you mean by "latency" as it relates to headphones? Its not a term I understand in the context of reviews here or in discussions of HP generally.

2. How do you define "transparent" in this context?

As far as my understanding of your questions vs measurements:

Yes, distortion obviously plays a part- distortion too high at the volumes you listen to and clearly it will be neither detailed or transparent. So has to be borne in mind and looked at.

As for Freq Response not being discussed- it has to be as there isnt much else to discuss apart from distortion and impedance. I think what you refer to as "detailed" may be due to excess treble- it brings with it a perception of detail but is at the expense of overall balanced sound. But then we get into the mire of psycho-acoustics , target curves etc as @Robin L was saying. The Beyers sound detailed because of the massive treble spike in the 990s and other models - but they compare terribly to target FR curves because of this and deficiencies in the bass and what we now know to be high distortion levels when pushed.

 

[steve59]

I find the answer to be convenient to a predetermined conclusion.

 

[Inner Space]

Specifically, what measurement (if any) will tell you something about the level of detail or transparency a headphone can provide?

"Detail" tends to mean low-level sounds meant to be heard, but often not, and "transparency" tends to mean the sensation of being able to hear them through everything else going on. Generally the associated measurements will be an elevated FR in the highs, possibly with a little distortion, which together make the "details" stand out more prominently.

 

[Robin L]

Couple of points I want to clarify with you first (and I am a million miles away from technically competent in this regard.

1. What do you mean by "latency" as it relates to headphones? Its not a term I understand in the context of reviews here or in discussions of HP generally.

2. How do you define "transparent" in this context?

As far as my understanding of your questions vs measurements:

Yes, distortion obviously plays a part- distortion too high at the volumes you listen to and clearly it will be neither detailed or transparent. So has to be borne in mind and looked at.

As for Freq Response not being discussed- it has to be as there isnt much else to discuss apart from distortion and impedance. I think what you refer to as "detailed" may be due to excess treble- it brings with it a perception of detail but is at the expense of overall balanced sound. But then we get into the mire of psycho-acoustics , target curves etc as @Robin L was saying. The Beyers sound detailed because of the massive treble spike in the 990s and other models - but they compare terribly to target FR curves because of this and deficiencies in the bass and what we now know to be high distortion levels when pushed.

I'd give the example of the Stax Lambda Pro headphones, used to own them [you would not believe how many 'phones I've owned, got 7 right now] They roll off the bass, have a spike in the mids. Gives the impression of the most detail [of what I've owned so far], but also could reasonably be accused of being 'etched' in character. The Drop 6XX have a similar character but are not 'etched', have a smoother response in the mids and the upper registers. In general, if a headphone is subjectively characterized as 'detailed' it usually points to exaggerated treble.

 

[Jimbob54]

I find the answer to be convenient to a predetermined conclusion.

Meaning what?

 

[bluefuzz]

Specifically, what measurement (if any) will tell you something about the level of detail or transparency a headphone can provide?

I think 'detail' and 'transparency' are precisely the kind of nebulous audiophile terms we are trying to avoid. I would imagine they stand for a slightly elevated treble response from ~3-5 kHz. Some headphones will inevitably react with your ear canal and pinna more euphoniously than others in that region.

 

[paulraphael]

I used to assume that detail in audio would be easily measured, much the way spatial resolution and modulation transfer are measured in optics. But I've figured out that most of what audiophiles talk about with detail is really perceptual—which is to say, it doesn't have anything to do with the actual detail present.

True detail, in terms of information theory, would probably be about signal-to-noise ratio, frequency response, and lack of distortion. But we know that we're not usually talking about this, because people often perceive music from vinyl LPs as being extremely detailed—and here we know that huge amounts of information are lost.

Is Inner Space suggested, some of the perception of detail comes from frequency response. Often a bright eq curve can unmask detail in the parts of the spectrum that we're most sensitive to. So speakers/headphones with a bump in the upper mids/lower treble can sound more detailed (even if they don't necessarily sound good). Same with a dip in the lower midrange. This added detail is not about added fidelity.

I believe it's been shown that distortion can sometimes increase people's perception of detail—which makes sense, because it adds higher frequency stuff to the signal. This also doesn't necessarily correlate with sounding good, and also obviously isn't about fidelity.

One place where perceptions of detail and actual information might correlate is with room acoustics. Well controlled early reflections will probably result in less smearing of transients. With headphones this could analogous to how well the cans work with your unique head and ear shape (as Robin L said).

 

[Robin L]

Here's an index of Amir's reviews of headphones. Click on any of them, you will be directed to the review for that headphone:

Headphone Review and Measurements Index | Audio Science Review (ASR) Forum

Also, Solderdude has his own site, has his own measurements, comments:

DIY-Audio-Heaven | Enjoy music through headphones on a low budget (wordpress.com)

Tyll Hertsens used to do headphone reviews and measurements, has a real-world attitude about the whole thing:

InnerFidelity | Stereophile.com

 

[Thomas_A]

I think this site is great and I appreciate the objective approach amir takes to review audio gear without the use of nebulous language. However I have owned and listened to many headphones and checking out the reviews here, I have problems relating my experience to the measurements.

Specifically, what measurement (if any) will tell you something about the level of detail or transparency a headphone can provide?

As an example, I own both the recently reviewed ATH-M50X (well I have the non-x version) and the DT990Pro. In the reviews the ATH-M50X shows better latency as well as better distortion measurements, but as much as I like the ATH-M50, in terms of transparency I think it is at least a league below the DT990Pro. The Beyerdynamic is far better in that regard.

How would I get that information from reading the measurements?

If it is a personal preference, you can't get any meaningful data from the measurements. Choosing between the DT770-880-990, I prefer the 880. 990 is way too bright for my taste. That said, among the Beyerdynamics, the DT150 with pads from the DT100 beats them all, IMO.

 

[hans132]

What is "more transparent" to one person is "shrill" to another, "dull" to a third.

I think 'detail' and 'transparency' are precisely the kind of nebulous audiophile terms we are trying to avoid. I would imagine they stand for a slightly elevated treble response from ~3-5 kHz. Some headphones will inevitably react with your ear canal and pinna more euphoniously than others in that region.

I think what you refer to as "detailed" may be due to excess treble- it brings with it a perception of detail but is at the expense of overall balanced sound.

I see what you mean. I am afraid there actually might be a language/subjectivity barrier within my question. I can relate to what you are arguing, when you say, that transparency is equated to a "bright" eq curve. However there seems to be more to it, as even when I eq the AT-M50, it seems to me that it still lacks detail, that I can hear with the DT990Pro. E.g. the HD650 and AT-M50 seem closer frequency curve wise than the M50 and DT990Pro, but still the HD650 seems to have far more detail than the M50.

1. What do you mean by "latency" as it relates to headphones? Its not a term I understand in the context of reviews here or in discussions of HP generally.

What I was referring to was group delay. Sorry for choosing the wrong term here.

So there won't be a single metric for "Transparency". Frequency response is a measurable metric, distortion is a measurable metric. "Transparency" is a subjective impression, not something measurable.

As for Freq Response not being discussed- it has to be as there isnt much else to discuss apart from distortion and impedance.

I was hoping that there may be a metric/aspect that I overlooked so far and maybe there isn't. However it is not scientific to talk about measurements, "just because we have them/because we don't have anything else". Just because there aren't any measurements/any good measurements doesn't mean that it is not a thing. You can't precisely measure things like intelligence for example. IQ tests have a lot of problems and there is even a lot of dispute about how to define something like intelligence in the first place. I guess "transparency" seems to fall into a similar category. On the other hand just because you _can_ measure distortion it doesn't necessitate it is relevant. E.g it is not relevant when you discuss volume.

Meaning what?

Yeah, what is the predetermined conclusion?

True detail, in terms of information theory, would probably be about signal-to-noise ratio, frequency response, and lack of distortion. But we know that we're not usually talking about this, because people often perceive music from vinyl LPs as being extremely detailed—and here we know that huge amounts of information are lost.

That makes a lot of sense to me. Sadly I never critically listened to vinyl, but I often find old Jazz recording much more revealing than modern day pop mixes, as latter aren't very dynamic in terms of volume. So it makes sense to me, that you could argue, that the modern recordings aren't lacking detail in the information theory sense, but instead it got greatly obfuscated.

Thanks everyone for trying to answer the question. While I am not super happy with the answer/conclusion, the issue seems to be rather with the question. I guess my question was far less clear than I initially assumed and at the very least I see how different factors could at least contribute to the perceived difference in detail, even tough I am sure I don't have a complete picture yet.

 

[Thomas_A]

An elevated HF region is often assoicated with more detail.

 

[raistlin65]

From your response I am not sure if you suggesting, that the ear canal affects the perceived detail. If you do, then I fail to see how that would be relevant? A less transparent speaker in a terrible room, would still sound less transparent than a more transparent speaker in the same room.

If by transparency, you are thinking of the accuracy of sound reproduction, I think you are underestimating the impact of the individual room.

It is the room and the speaker's characteristics together that affect performance. Speaker A might be technically better than Speaker B based on anechoic measurements. Speaker A could be better in Room X, yet Speaker B could be better in Room Y.

 

[Robin L]

I see what you mean. I am afraid there actually might be a language/subjectivity barrier within my question. I can relate to what you are arguing, when you say, that transparency is equated to a "bright" eq curve. However there seems to be more to it, as even when I eq the AT-M50, it seems to me that it still lacks detail, that I can hear with the DT990Pro. E.g. the HD650 and AT-M50 seem closer frequency curve wise than the M50 and DT990Pro, but still the HD650 seems to have far more detail than the M50.



What I was referring to was group delay. Sorry for choosing the wrong term here.



I was hoping that there may be a metric/aspect that I overlooked so far and maybe there isn't. However it is not scientific to talk about measurements, "just because we have them/because we don't have anything else". Just because there aren't any measurements/any good measurements doesn't mean that it is not a thing. You can't precisely measure things like intelligence for example. IQ tests have a lot of problems and there is even a lot of dispute about how to define something like intelligence in the first place. I guess "transparency" seems to fall into a similar category. On the other hand just because you _can_ measure distortion it doesn't necessitate it is relevant. E.g it is not relevant when you discuss volume.


Yeah, what is the predetermined conclusion?


That makes a lot of sense to me. Sadly I never critically listened to vinyl, but I often find old Jazz recording much more revealing than modern day pop mixes, as latter aren't very dynamic in terms of volume. So it makes sense to me, that you could argue, that the modern recordings aren't lacking detail in the information theory sense, but instead it got greatly obfuscated.

Thanks everyone for trying to answer the question. While I am not super happy with the answer/conclusion, the issue seems to be rather with the question. I guess my question was far less clear than I initially assumed and at the very least I see how different factors could at least contribute to the perceived difference in detail, even tough I am sure I don't have a complete picture yet.

I'll throw you a subjective curveball. While the numbers for frequency response, distortion and group delay just might add up to a signature of resonant behavior, it doesn't completely cover the subjective impression of resonant behavior. A good example of this is the Philips Fidelio X2 HR headphones [Amir and Solderdude reviewed them, had different subjective impressions]. What I hear is a little fuzzier, less "transparent" than the AKG K371 or 6XX 'phones. However, it is also more spacious, seems to have the image float [a little] outside the head, unlike the closed-in sound of the other two 'phones. There's a little bit of shrillness up top, things seem slightly unfocused. This works great with my Duke Ellington recordings. In any case, I "know" I'm hearing some weird resonant activity, probably related to the resonant behavior that enables these open-backed headphones to dig into the bass. I haven't torn open what's inside these headphones, but I'll bet the moving mass of the driver is greater than the moving mass of the Stax earspeakers. And then I recall the Stax 'phones, and how the drivers consisted of thin, light plastic membranes, how little mass was involved, I get the impression of speed and transparency just looking at them. I would pontificate [without a technical background, just an accumulation of subjective responses] that we can hear the "sound" of materials that are right next to our ear a lot easier than hearing them from across the room.

 

[Jimbob54]

I see what you mean. I am afraid there actually might be a language/subjectivity barrier within my question. I can relate to what you are arguing, when you say, that transparency is equated to a "bright" eq curve. However there seems to be more to it, as even when I eq the AT-M50, it seems to me that it still lacks detail, that I can hear with the DT990Pro. E.g. the HD650 and AT-M50 seem closer frequency curve wise than the M50 and DT990Pro, but still the HD650 seems to have far more detail than the M50.



What I was referring to was group delay. Sorry for choosing the wrong term here.



I was hoping that there may be a metric/aspect that I overlooked so far and maybe there isn't. However it is not scientific to talk about measurements, "just because we have them/because we don't have anything else". Just because there aren't any measurements/any good measurements doesn't mean that it is not a thing. You can't precisely measure things like intelligence for example. IQ tests have a lot of problems and there is even a lot of dispute about how to define something like intelligence in the first place. I guess "transparency" seems to fall into a similar category. On the other hand just because you _can_ measure distortion it doesn't necessitate it is relevant. E.g it is not relevant when you discuss volume.


Yeah, what is the predetermined conclusion?


That makes a lot of sense to me. Sadly I never critically listened to vinyl, but I often find old Jazz recording much more revealing than modern day pop mixes, as latter aren't very dynamic in terms of volume. So it makes sense to me, that you could argue, that the modern recordings aren't lacking detail in the information theory sense, but instead it got greatly obfuscated.

Thanks everyone for trying to answer the question. While I am not super happy with the answer/conclusion, the issue seems to be rather with the question. I guess my question was far less clear than I initially assumed and at the very least I see how different factors could at least contribute to the perceived difference in detail, even tough I am sure I don't have a complete picture yet.

For what it is worth, I know what you mean about detail- I have a few sets of decent phones and all sound different despite having similar EQs dialled in. And some definitely seem to have more detail, so I'm not a purist saying get the FR the same and they sound the same. But I am happy that the bigger differences in presentation come down to FR and distortion, as well as angle of drivers etc. And I find it impossible to rule out the impact comfort has on ones perception of sound . I truly think uncomfortable phones will "sound" worse, because we are human, not because sound waves hitting our ears.

 

[3125b]

However, it is also more spacious, seems to have the image float [a little] outside the head, unlike the closed-in sound of the other two 'phones

I'm pretty sure by now that that has a lot to do with the position, especially angle, of the driver relative to the ear.
Focal writes about that in their white paper for the Elear and Utopia, and that aligns well with my own subjective impression from the headphones I know. The X2HR has angled drivers, K371 and 6XX don't.
The Elear, wich has the drivers mounted fairly far in front of the ear at a fairly acute angle and is extremely open (you can see right through), works best for me out of all my headphones.
But even then, that can't be everything there is to it. I own both a HD 555 and a 569, they have the exact same construction with angled drivers, only the 569 is a closed back while the 555 is open, and of course they have different drivers. Funnily enough, the 569 works really well in terms of sound stage and imaging for me, while the open back 555 just doesn't.
And then there are the AKGs. The K702 doesn't have angled drivers, it only has slightly angled pads, not a lot yet still works very good.
Unfortunately, getting an objective position on that hypothesis is unlikely, because how would one measure that?

@topic:
The level of Detail depends mostly on frequency response. Of course distortion should be low as well, though there is a distinction to be made between "subconsciously changing the overall perception" and "plainly audible".
What I found is that I enjoy headphones with a fairly even FR more than ones with a lot of peaks and dips.
Many people with little experience seem to regrad headphones with a lot of upper highs (like most Beyerdynamics) as especially transparent, I (and many more experienced people seem to as well) just can't stand it anymore.