Can we measure detail retrieval?

[ZENERGiA]

Hey guys!

Thanks to you, I got my feet wet in the audio world, and I am having good sound quality, without spending a fortune. My current favorite headphone (which I own) is the k712 pro, which I quite enjoy (tonality wise it might not be the best, but I like the somewhat "airy" sound with the "wide soundstage").

Now what I am really interested in, is the following: is there a way to measure detail retrieval, or in general the "correctness" of the sound (not speaking tonality wise)? I am sure there are a lot of headphones out there, that are tonally correct, but just do not have good detail retrieval, or the timbre is not good quality, etc.

How do I know (from a scientific point of view) that the detail retrieval of a headphone, and the timbre qualities are good, the headphone will sound "airy", etc.?

Thank you for your responses in advance!

 

[MayaTlab]

How do I know (from a scientific point of view)

Well from a scientific point of view, terms that have yet to receive an operational definition are difficult to assess . I'm not aware of any operational definition for "detail retrieval" as far as headphones are concerned, and I'm afraid that a lot of the debate that occurs around the terms often used by audiophiles to describe sound has more to do with research into semantics than acoustics.

What we do know, however, is that quite a few of the measurable aspects of sound have a threshold for audibility, and that with current headphones the one remaining variable that seems to vary between them far more than this threshold of audibility is frequency response. The main problem being that frequency response not only varies across headphones, but also listeners because of how headphones interact with your anatomy in various ways - and it might actually be desirable if it matches the way your anatomy modulates the FR curve of natural sound sources (look into HRTF), but it's unlikely that headphones interact with your ears in such a way. As long as FR curve at one's eardrum remains an uncontrolled variable (and as said it might actually be desirable that it varies, but in a very specific way for each listener), it's difficult to attribute to another factor the cause of what you're hearing. Therefore it's difficult to measure whatever "detail retrieval" would be if operationally defined without taking into account frequency response.

In short, IMO as long as FR curve remains an uncontrolled variable which undesirable variability far exceeds the threshold of audibility, it's likely that whatever "detail retrieval" means to various people is related one way or another to frequency response at one's own eardrum.

 

[MDT]

To emphasize the point made by MayaTlab, can you first please define "detail retrieval"? The first thing any scientist must do when tackling a problem, is to clearly define it.

 

[mkawa]

we need to define it rigorously first. it's one of those ambiguous audiophile words that means nothing specific as far as measurability and known properties of transducers is concerned

 

[ZENERGiA]

we need to define it rigorously first. it's one of those ambiguous audiophile words that means nothing specific as far as measurability and known properties of transducers is concerned

By detail retrieval I would mean the little nuances in the recording, for example when the singer inhales air, in the beginning seconds of the track, before singing (Sting - My funny Valentine) or when their lips are opened, before they start singing, or a little studio noise, etc. I would also be interested in the correctness of the timbre, and in general how lifelike the sound is percieved by the listener, when wearing the particular headphone.

 

[Jimbob54]

By detail retrieval I would mean the little nuances in the recording, for example when the singer inhales air, in the beginning seconds of the track, before singing (Sting - My funny Valentine) or when their lips are opened, before they start singing, or a little studio noise, etc. I would also be interested in the correctness of the timbre, and in general how lifelike the sound is percieved by the listener, when wearing the particular headphone.

Depends on a few things I guess, not least of which is the attentiveness of the listener at the time and the overall volume and quality of recording too.

But leaving those things to one side- frequency response will play a large part. A headphone that boosts the frequencies of the lip smack/ breath etc is more likely to sound detailed in those frequencies- but by extension must be accentuating some other frequency less. One of the reasons various Beyerdynamics are famed for their impression of detail is their considerable exaggeration of treble frequencies. This may not be pleasant for some listeners and makes them far less than "neutral".

Thats not to say there arent other factors involved too. Headphones , the understanding of how they interact with us and we perceive them seems like far from a done deal from a scientific perspective from what I have read here and elsewhere.

 

[Zensō]

A headphone that boosts the frequencies of the lip smack/ breath etc is more likely to sound detailed in those frequencies- but by extension must be accentuating some other frequency less. One of the reasons various Beyerdynamics are famed for their impression of detail is their considerable exaggeration of treble frequencies. This may not be pleasant for some listeners and makes them far less than "neutral".

I agree with this. “Detail retrieval“ as described by audiophiles seems to be related to certain elevated treble frequencies. Most of the headphones described as having great detail retrieval by some, are also described as being overly bright by others (HD800 and DT1990 Pro come to mind).

 

[Duckeenie]

Depends on a few things I guess, not least of which is the attentiveness of the listener at the time and the overall volume and quality of recording too.

But leaving those things to one side- frequency response will play a large part. A headphone that boosts the frequencies of the lip smack/ breath etc is more likely to sound detailed in those frequencies- but by extension must be accentuating some other frequency less. One of the reasons various Beyerdynamics are famed for their impression of detail is their considerable exaggeration of treble frequencies. This may not be pleasant for some listeners and makes them far less than "neutral".

Thats not to say there arent other factors involved too. Headphones , the understanding of how they interact with us and we perceive them seems like far from a done deal from a scientific perspective from what I have read here and elsewhere.

A good answer that respects the OP. I'd vote you up more than once if I could.

 

[RHO]

Could it be that non-linearity plays a roll in all this. Headphones do have mechanical parts in the system responsible for the transfer of the signal. So you could measure a perfectly smooth FR, fitting perfectly to the target @94dB. But maybe @60dB it deviates from the target significantly. This would show that small details in the sound could be exaggerated or smoothed out.
...just thinking out loud...

 

[Robin L]

By detail retrieval I would mean the little nuances in the recording, for example when the singer inhales air, in the beginning seconds of the track, before singing (Sting - My funny Valentine) or when their lips are opened, before they start singing, or a little studio noise, etc. I would also be interested in the correctness of the timbre, and in general how lifelike the sound is percieved by the listener, when wearing the particular headphone.

Those sorts of details might be indicating a distortion of some sort, an exaggeration of certain frequencies that pop out of the mix when they were not intended to be heard by the producers/performers of a given recording. My Stax earspeakers did that, made things like feet shuffling in an orchestra recording audible when it was not actually intended to be heard.

I'd still go for: noise floor has to be low, distortion has to be low, frequency response has to be even. The Drop 6XX headphones sound similar to the Stax 'phones, save that the exaggerated elements don't pop from the mix the way they did on the Stax 'phones. One difference, one that might be more important ultimately, is having clear, distinct pitches with percussion. Usually, there's a sound more like evenly distributed noise on a snare than an actual pitch. That was the first thing I noticed when I got the Drop 6XX/Topping E/L 30 combo a few months ago, hearing the pitches from drum kits instead of a vague noise. But having low-level details pop from a mix is probably an indication of something being heard that wasn't intended to be heard in the first place.

 

[mkawa]

A good answer that respects the OP. I'd vote you up more than once if I could.

interesting!!

 

[MDT]

By detail retrieval I would mean the little nuances in the recording, for example when the singer inhales air, in the beginning seconds of the track, before singing (Sting - My funny Valentine) or when their lips are opened, before they start singing, or a little studio noise, etc.

Whether these can be heard depends on the a number of factors, essentially: noise, frequency response and distortion. If the sound is lower than the noise level, you won't hear it. If the frequency range of those sounds are in a region where the frequency response is suppressed, they will be quieter and if they are quiet and you need to turn up the volume, distortion becomes relevant. There's no single measure that takes all this into account because the response is not uniform for speakers or headphones whereas for DACs and Amps with a flat response this is trivial.

I would also mention that "detail retrieval" will vary between tracks for the reasons mentioned above. Some tracks will better match some headphones than others. The ideal would be a headphone that has a completely neutral profile for your own hearing.

 

[Wes]

Let's make this simple: we do a blind test on a listening panel and simply ask them whether A or B has more detail. We need not define detail for them, as they will have some preconception about the defn. If there is a sig. difference, we can then proceed to other tests or analyses.

 

[paulraphael]

I agree with this. “Detail retrieval“ as described by audiophiles seems to be related to certain elevated treble frequencies. Most of the headphones described as having great detail retrieval by some, are also described as being overly bright by others (HD800 and DT1990 Pro come to mind).

You can demonstrate this with eq. If you know what to do, you manufacture detail where before there had been less. Which demonstrates that it's mostly a perceptual phenomenon ... not what audiophiles think it is.

Cutting some upper bass / lower midrange frequencies has a similar effect. It's why we perceive "detail" and "warmth" as being in opposition.

I think there are some other things that effect the sense of detail, including room acoustics. Reflections of a certain type can blur detail. In this case it's probably actual detail loss and not just a perceptual phenomenon.

 

[xaviescacs]

Besides FR, it's not a time domain issue? I mean, detail to me sounds like I'm able to distinguish between two "events", either in frequency or in time.

 

[Martin]

By detail retrieval I would mean the little nuances in the recording, for example when the singer inhales air, in the beginning seconds of the track, before singing (Sting - My funny Valentine) or when their lips are opened, before they start singing, or a little studio noise, etc. I would also be interested in the correctness of the timbre, and in general how lifelike the sound is percieved by the listener, when wearing the particular headphone.

Depends on a few things I guess, not least of which is the attentiveness of the listener at the time and the overall volume and quality of recording too.

But leaving those things to one side- frequency response will play a large part. A headphone that boosts the frequencies of the lip smack/ breath etc is more likely to sound detailed in those frequencies- but by extension must be accentuating some other frequency less. One of the reasons various Beyerdynamics are famed for their impression of detail is their considerable exaggeration of treble frequencies. This may not be pleasant for some listeners and makes them far less than "neutral".

Thats not to say there arent other factors involved too. Headphones , the understanding of how they interact with us and we perceive them seems like far from a done deal from a scientific perspective from what I have read here and elsewhere.

When I completed my subjective comparison of four different higher end headphones one of the things I found I was evaluating was "detail retrieval." What I found was my impression of "detail retrieval" was largely dependent upon my attentiveness to the recording. When I found examples of these details - breaths, studio noises, guitar string brushes or instrument body sounds, etc. - while wearing one headphone I could always hear it when I listened for it on another headphone. Some were better at emphasizing it than others - frequency response - but none masked it completely. For me I found my attentiveness was by far the greatest differentiator between if I heard detail or not.

Martin

 

[paulraphael]

When I found examples of these details - breaths, studio noises, guitar string brushes or instrument body sounds, etc. - while wearing one headphone I could always hear it when I listened for it on another headphone. Some were better at emphasizing it than others - frequency response - but none masked it completely.

This sounds like a description of sharpness in images. It's an idea that people often conflate with resolution, but is significantly different. Resolution is a bit tricky to measure; it's basically a measure of the finest details that can be differentiated, usually expressed in line-pairs/mm. It would be similar to an audio frequency response measure of where the FR drops below -70db (or something like that ... a point at which most people find it barely audible).

But sharpness is a subjective quality, and it correlates with the amount of contrast at a specific range of spatial frequencies. An image that has high contrast in the range of 1 lp/mm and 5 lp/mm will look sharp. You'd be able to see the detail in those spatial frequency ranges at lower contrasts, but they'd look hazy.

Interestingly, sharpening filters essentially boost the amplitude at certain spatial frequencies—ones we'd think of as mid- or lower-treble.