What's Left In Speaker Design To Reduce Distortion/Increase Detail Retrieval?

[fineMen]

IMHO, the best speaker designs would have flat frequency response, if, and only if, the recordings we buy were any way near flat.

Plus what you say about intermodulation from amplitude or Doppler distortion.

The common man won't effortlessly get what he cannot accept. I gave up.

Recording is an artform in itself with so many decisions taken in every aspect, of whom each would make it 'sound' completely differnt. Which is the purpose of the decision making. Who's the judge on good sound? The sound engineers and mixers and producers and maybe the musicians or even the composer. What's the reference? The monitors in the studio, actually. Is the esthetic judgement standardized? No. Is the monitor standardized? No. So, where is my 'original' that was promised to me by the industry? Better don't ask but rather expect there's an essence taht is revealed by high tier stereo.

Doppler's distortion, or IM in general is a multidimensional measurement not covered by advertizing and so called digital/paper test magazines. It asks for re-evaluation of some technologies in general and shockingly doesn't provide an easy ranking of products. Better don't touch.

After these two topics were dismissed as irrelevant we are still left asking what criteria to optmize for 'detail retrieval'. My final question was, what kind of a detail is retrieved by whom in the signal chain. Is it the tech, or is it the listener's imagination? Where does the damand come from, missing what exactly? Is the expectation legit? (HiFi kitsh often adds pseudo detail, like screeching guitar strings, breathing noises and what have you.)

 

[Pudik]

That is an excellent example of the anthesis of what Audio Science Review is all about. Vague, meaningless, amorphous descriptions instead of measurements and reality.

Anthesis? You mean antithesis, don't you...

 

[IPunchCholla]

Is the monitor standardized?

The monitor isn’t standardized, but many of them, even less expensive ones are +-3dB from 50 to 80 all the way up to 20000 hZ. And I have never met a mixing engineer who only mixed to a single pair of monitors. Most listen on a mix, including intentionally designed crappy ones and headphones in order to extrapolate an ideal mix. It is amazing how well humans can do this. If I listen to the same song on various playback devices I can quickly construct an “ideal” version of that song and then tell you how each device is altering it, even if none of the devices approach anywhere near the ideal.

It is the same mechanism that let’s us pick out an individuals voice, even though our voices vary from minute to minute, year to year and are heard in incredibly varying acoustic situations.

 

[fineMen]

..., including intentionally designed crappy ones and headphones ...

O/k, I often asked when this argument was used again, what a crappy speaker is. It is kind of a tool for the engineer (?) but not well defined.

... in order to extrapolate an ideal mix. It is amazing how well humans can do this. If I listen to the same song on various playback devices I can quickly construct an “ideal” version of that song and then tell you how each device is altering it, even if none of the devices approach anywhere near the ideal.

In turn that would mean that no mix is as good as it could be if designed to be listened to using just the 'ideal'.
And vice versa, what's the benefit of longing for the 'ideal' then on the consumer side?

It is the same mechanism that let’s us pick out an individuals voice, even though our voices vary from minute to minute, year to year and are heard in incredibly varying acoustic situations.

The 'essence'? I don't think this analogy fits. First a voice's signature is widely maintained even using crappy speakers or having it in a failed mix (if there is such a thing). Or could you distort a voice using every-day methods so far that it would be identified as another?

As a sound engineer / mixer type you're more of an artist using tech/ for expressing something. What is it? Your view, turned around for somebody else to see through your .. ears. Focus on the soloist, or the background choir this time, you know what I mean. Together with the musicians you offer to put an enjoyable meaning into sound. (Not triggering the primitive stone age's escape reflex, as many sophisticated audiophiles quote it with happiness.) You're as much an artist as the musician, admit it finally ;-) C'mon, even a photographer thinks of her as an artist ...

So meeting the artist, what "detail" (see o/p title) are we talking here?

 

[Travis]

There is no 'myth of flat recordings'. That's nonsense.

And a lot of the rest reads like ancient audiophile nonsense too.

On the distortion issue, If he’s quoting Heyser and Keele it isn’t myth, especially Heyser. You can get Heyser’s papers from AES for free.

 

[IPunchCholla]

O/k, I often asked when this argument was used again, what a crappy speaker is. It is kind of a tool for the engineer (?) but not well defined.

Here’s one: Mixcube

In turn that would mean that no mix is as good as it could be if designed to be listened to using just the 'ideal'.
And vice versa, what's the benefit of longing for the 'ideal' then on the consumer side?

This is a misunderstanding. Mixing engineers are mixing to make the best sound under the conditions they believe music will be listened under. There is often more than one mix based on an understanding of those conditions. LPs are sometimes mixed differently than digital for pop music. Classical has different assumptions about the listening space than pop. I never said anything about longing for the ideal. I think I would agree with you that it doesn’t exist.

One can distort a voice past recognition using what are now every day tools. But just because something can be changed beyond recognition doesn’t negate the fact that humans are incredibly capable of creating a mental model of a sound by listening to exemplars of that sound in varying contexts.

As for detail retrieval, I also don’t know exactly what that means.

 

[fineMen]

Here’s one: Mixcube

Yep, they advertize the product with 'pair of monitors to replicate the wide range of “bass-challenged” devices such as car' but also praising *a more smooth, open, and transparent driver ... ... ...'. Why not simulate less bass with an equalizer for the main monitors then?
This basically isn't engineering, it is an artist's view, looking at it from the result. Established tradition, the lead standard, not a science based tool.

This is a misunderstanding. Mixing engineers are mixing to make the best sound under the conditions they believe music will be listened under.

As they should. But what is 'best sound'? I leave it at that.

As for detail retrieval, I also don’t know exactly what that means.

Agreed again.

 

[MattHooper]

As for detail retrieval, I also don’t know exactly what that means.

I'm curious: Would that go for visual images as well? If someone said the image from a Blu Ray displayed more detail than the same image on a DVD, would you not
know what they were talking about?

 

[garyrc]

There is no 'myth of flat recordings'. That's nonsense.

And a lot of the rest reads like ancient audiophile nonsense too.

Thank you so much for your sensitive reply.

By the "myth of flat recordings" I meant:

1. The belief that if one sets the tone controls for "flat" [or uses a preamp with no tone controls] one will get flat sound i.e., the music will sound more like what the engineers heard than if you manipulate the tone controls [adjusting by ear], and/or use an appropriate ISO 226-2023 curve unless you play back at control room SPL.
2. The belief that with the tone controls set on "flat" you will hear "flat" in your room, from a given (not flat) recording, which is likely to have been monitored in a control room rather different than your listening room, over speakers rather different than yours, etc.

 

[IPunchCholla]

I'm curious: Would that go for visual images as well? If someone said the image from a Blu Ray displayed more detail than the same image on a DVD, would you not
know what they were talking about?

I would absolutely know what they meant. I teach photography at the university level and can explain where in the reproduction chain loss of detail can occur and why for several types of digital reproduction and film, from the glass on the camera to the print.

but I am still not sure what people are referring to with audio. This is because there is huge confusion around the relationship of transients to FR, a basic muddiness surrounding human sensitivity to distortion in relation to masking effects, and a certain muddiness around speaker room interaction and DSP. Much of this muddiness is a result of me being a newbie at this stuff, but a lot of it is also a general imprecision of people’s language around these things.

For example, I’ve been mixing a song lately, the violins come in very quietly before a massive distorted chord is played. I can’t hear the violins before the chord on my monitors, but I can on my headphones.Over speakers, the room noise is too high, unless I go up to uncomfortable listening levels. Is that a speaker detail issue?

 

[MattHooper]

I would absolutely know what they meant. I teach photography at the university level and can explain where in the reproduction chain loss of detail can occur and why for several types of digital reproduction and film, from the glass on the camera to the print.

Cool!

(I majored in film and photography, myself...)

but I am still not sure what people are referring to with audio. This is because there is huge confusion around the relationship of transients to FR, a basic muddiness surrounding human sensitivity to distortion in relation to masking effects, and a certain muddiness around speaker room interaction and DSP. Much of this muddiness is a result of me being a newbie at this stuff, but a lot of it is also a general imprecision of people’s language around these things.

Sure. For the sake of zeroing in on my point, I think we need to separate the sound from the cause. We can percieve that there is "more detail" in one presentation, less in another. Whatever the particular cause might be for the difference is another thing.

Also, I agree someone can just throw out the term "more detail" without being specific. But, one can just be more specific. You've given me a specific picture of what you are hearing, which helps.

For example, I’ve been mixing a song lately, the violins come in very quietly before a massive distorted chord is played. I can’t hear the violins before the chord on my monitors, but I can on my headphones.Over speakers, the room noise is too high, unless I go up to uncomfortable listening levels. Is that a speaker detail issue?

I'd say you are indeed describing a difference in audible detail. As to the cause, I wouldn't be in a position to know. But if you had two different speakers, in the same room conditions, and you heard that kind if difference in detail, I'd think it would point to one speaker being less resolving of detail.

Another example: I use two different preamps, a tube preamp and a Benchmark preamp. The Benchmark preamp seems to resolve slightly more recorded detail. (I've blind tested between them btw). For instance, if there is a distant instrument at the "back" of the mix, with a very subtle amount of reverb, that light halo of reverb will be more easily audible on the Benchmark amp. In music production there is often used the terms "dry" and "wet" - where "dry" is less reverb, and "wet" is increased reverb. The way the Benchmark amp resolves the most subtle amounts and trails of reverb and recorded acoustics in recordings, the difference between the Benchmark and tube preamp presentation is like having a very fine "reverb" dial, where it is very slightly dialed to "more wet sounding" for many tracks, and a bit "more dry sounding," like I've just slightly dialed back the reverb, on the tube pre-amp. Of course, I'm talking about *very* subtle differences in detail in that case.

 

[fineMen]

Cool!

(I majored in film and photography, myself...)

Question regarding detail in contemporary photography: why is it that the reconstruction of the Bayer matrix pixels into an image is done using gradients and stuff like that, but not using neural networks of AI fame? Especially as the neural network resembles not 'brain' but the pre-processing in the eye's retina, that has also pixels, but chaotically distributed.

Obviously optics isn't a valid model for audio.

Finally I propose to see a lack of detail as an asset. It keeps you interested in the very same recordings for longer. One detail at one day, the other for the next. Imagine to reveal all details at the same time, as you would (not?) do in a real concert. Detail perception is in the listener's realm, not the machines's property.

If there's still a problem with that, at least on my side I would love to pull the term 'detail' out of the audiophile's territory first. Is has to relate to something other than subjective excitement to gain some meaning. Same applies to 'stereo imaging'. Funny terms of no further use so far.

(And again a recording is understood as a 'reproduction' of some invariable indescriptive essence, whilst all the artful decision making during the recording process as a whole is dismissed.)

 

[IPunchCholla]

Question regarding detail in contemporary photography: why is it that the reconstruction of the Bayer matrix pixels into an image is done using gradients and stuff like that, but not using neural networks of AI fame? Especially as the neural network resembles not 'brain' but the pre-processing in the eye's retina, that has also pixels, but chaotically distributed.

Historical accident. Bayern’s work was patented in 1976 and put in to practice around the same time. Neural networks and the associated processing power to run them, didn’t exist. Our processes of demosaicing were highly developed at the point neural nets on camera became feasible. While they are feasible on computers where you can process RAW files, even Adobe, which is incorporating a lot of ML into their tools, didn’t use the tech in their raw converter last I looked. Most likely because the benefit would be so small at this point.

Perhaps if we went back and figured out how to make chips with random arrangements (in both spacing and pattern) of RGB sensors, a neural net could make a process that would be “better” than our current one. But it wouldn’t be significantly better and likely not even visibly better, except to weirdo like me who will look at reproductions under a loop to check for errors.

I think some reasonable analogies can be made about visual reproduction vs audio, but the sampling methods being so different does cause problems.

Where your question does get me thinking is in the notion of trying to unthink current transducer design and consider if there are ways using current technology to create a totally different type of transducer that 1. Is far more efficient, and 2. Less prone to current sources of distortion without introducing to many new ones.

 

[fineMen]

... to check for errors.

Understood.

I think some reasonable analogies can be made about visual reproduction vs audio, but the sampling methods being so different does cause problems.

Frankly spoken, I issue the notion on what a recording is quite often, because I can no more accept the attempt to check a speaker's quality in having it compared to a live session. But that may not affect you, neither my discomfort nor the comparison.

The sampling you refer to is, for me, not the digitization, but the - ha - picture's composition. With audio there are so many more degrees of freedom compared to photography, that it is more like painting. It is a completely dfferent dimension. But again, I won't get through. Never mind

 

[MattHooper]

Finally I propose to see a lack of detail as an asset. It keeps you interested in the very same recordings for longer. One detail at one day, the other for the next. Imagine to reveal all details at the same time, as you would (not?) do in a real concert. Detail perception is in the listener's realm, not the machines's property.

That is a misleadingly simplistic analysis.

Of course the end result of audio is our subjective perception, but the idea that perception of detail has *only* to do with us, and not to the physical thing we are perceiving, is untenable.

I do sound design for film/TV, and all day long I'm putting together layers of sounds, specifics and background tracks, and adjusting the volume and cutting the sound so that some details become more prominent, others less. These are physical, objective changes to the sound which will affect the perception of the sound, in terms of how the details are heard. Likewise, recordings vary and we constantly have to manipulate recordings to remove, reduce or enhance details in the recordings, whether it's dialogue recordings or sound effects. In the same way it should be obvious that anything from the recording quality to the quality of reproduction (e.g. competence of a sound system) affect detail perception.

 

[fineMen]

That is a misleadingly simplistic analysis.

Alas, I cannot agree with this. Arguments were given above.

 

[MattHooper]

Alas, I cannot agree with this.

Then you should be able to tell us why.

Arguments were given above.

Except the post you just replied to explained why your argument didn't make sense.

 

[fineMen]

Except the post you just replied to explained why your argument didn't make sense.

Matt, I feel tempted to revisit the famous movie "Yaws". All the detail in constructing a threat from below, the huge water body unknown to air breathing entities, the very first 5 -- sorry the first 120, minutes are genious. I won't make fun of anyone, but I mean what I said. Please define 'detail' as you are the o/p.

 

[MattHooper]

Matt, I feel tempted to revisit the famous movie "Yaws". All the detail in constructing a threat from below, the huge water body unknown to air breathing entities, the very first 5 -- sorry the first 120, minutes are genious. I won't make fun of anyone, but I mean what I said. Please define 'detail' as you are the o/p.

Ok, that's incomprehensible to me. Are you the fellow who denies stereo imaging as well? If so, maybe let's not bother with this.

 

[fineMen]

Ok, that's incomprehensible to me. Are you the fellow who denies stereo imaging as well? If so, maybe let's not bother with this.

As your alias clearly refers to the 70s movie "Yaws", which for sure is plain genius.

What I'm after is that
a) nobody expects a movie to be 'virtual reality', everybody accepts some effort to understand what is depicted
b) some movies are worth to be seen more often
c) only later one is able to identify the making of the movie, all those details, cues, hints whatever word is appropriate

My point is, the 'detail retrieval' is not in the technology of either the recording or the replay, it is in the listeners imagination. His/her effort to identify unexpected properties. What else is a 'detail'? It'll better be defined (by you as the o/p) before a discussion goes peculiar ways.

Theory shall allow to make provable predictions (hypothesis)! Prediction: Once an audiophile is believably told that a 'system' is quite expensive, he will in short time be able to identify 'details' in a recording, because he is willing, due to the expenses, to put some effort in´to mindful listening.