Measurements & Sound Quality

[fas42]

Good thread topic ... the obvious one for me is the transition into what I call competent sound - subjectively, this can be quite a dramatic difference, but objectively all that has happened is that the level of some distortion or noise anomaly is just a bit less than that prior to the transition. I believe any sort of conventional measurement, trying to pinpoint the key difference, would show very little of interest - specialist test signals most likely would need to be developed, to make it easier to register the change taking place.

 

"Sense of space" or "real instruments playing in real space" are part of the package of competent playback - I wouldn't see any point in trying to separate out those, as individually 'measurable'.

[fas42]

What most people are quite unaware of is the dramatic improvement in the subjective presentation when the objective quality passes a certain point. So an improvement of 1% in some measurable parameter may result in a subjective improvement of 200%, to pull a figure completely out of the air - if the difference is a rig sounding like Yet Another Hifi, versus being 'convincing', then for some people large numbers like that may seem right.

 

Because the system is not actually "better". That 1% is actually transitioning from 99 to to the necessary 100 "points of competence", because the last key degrading "point of weakness" has been resolved.

 

Why this happens is because the human hearing system demands a certain standard of accuracy before allowing itself to be fooled - anything below the standard is immediately rejected; anything above permits an illusion to be sustained.

[fas42]

5 hours ago, iaval said:

A good exercise might be to learn how a particular distortion sounds like (in exaggerated simulated scenario), and use that knowledge to confirm which measurement to use to confirm it. 

 

Very true, but some of the worst offenders can't be easily 'canned' - that is, switched on and off to one's hearing, in a neat, controllable package.

 

Just take all the varieties of digititus - fans of analogue well know the irksome qualities of digital sound "not being quite right" - examples being a dull, listless presentation; or detail being lost in a "black hole"; or a persistent, disturbing edginess. These are all valid distortion artifacts; but how does one create a nicely predictable sample of each of these anomalies?

[fas42]

10 minutes ago, iaval said:

Isolating the elements is vital for proper identification and study. If those effects are real, there should be some reliable method of confirmation.

Once isolated, then try to figure out what's going on. 

 

Agreed. Unfortunately, in the thirty odd years since we've have directly used digitally stored source in playback situations there has been scant interest in precisely assigning various causes to the "ills" of digital sound. At one point a catch-all misbehaviour was assigned: jitter - but this is a poor choice, IMO.

 

IME lack of engineering robustness at a system level is a key factor, as evidenced by the fact that interatively introduced strengthening and workarounds, in the various parts of the chain, resolve these SQ issues.

[fas42]

14 hours ago, iaval said:

Usually not without introducing other problems. 

 

Some would believe this, but I have found it to be otherwise. The main 'trick' is to precisely focus on aspects of the sound which are faulty - the one word that should never enter the discussion is "better"; at no time does one do anything to make the sound "better" - as I sometimes say, one Subtracts Badness, it's never an exercise in Adding Goodness; though the latter will be a by-product, automatically, in the subjective sense.

 

Quote

I'm a bit unclear on this - digital in this context is meant as PCM? 

The biggest problem with the CD era music IMHO is that dreaded loudness war, and the fact that many artists can't even render lossless track out of their DAWs and therefore supply labels with MP3s. 

 

Any mechanism where the music source can be treated as computer processable data.

 

The loudness war is due in part to the failure of audio systems to convey the vibrancy and energy of live sounds. So they substitute loudness as an easy "fix" - a bad cook adds too much salt or sugar to the dish to "give it some punch", and this is the deplorable state of much of the current recording production.

 

Note that one has to be careful not to confuse dynamics compression with data compression - entirely distinct processes, with very different subjective results.

[fas42]

9 hours ago, jabbr said:

So really the question should be: what in our living room is lacking that a live concert has? Send that signal to the brain  

 

Simple answer: a competent playback system . Yes, very trite, but doesn't stop it being true - if one has spent years fiddling with rigs that can be on either side of the necessay degree of optimisation to "send out the right signals to the brain", depending upon everything, then it's a remarkably straightforward question to deal with.

 

If the living space could magically detect and null out the distortion artifacts that the brain is reacting to then that's another solution. And room treatments are a well worn, sortof method of trying to do this. But the best remedy is to just get the playback chain to be on its best behaviour - and then one's brain does the rest ..

[fas42]

6 hours ago, NOMBEDES said:

Can you plot human hearing on a graph?  A bell curve?

Do any two people hear something as complex as music in the same way?  

 

So how can we agree on any relationship between measurements and sound quality?

 

Put 10 audiophiles in a room and you may get 7 opinions regarding sound quality regardless of measurements.  (and who knows if measurements are accurate anyway?)

 

 

 

Best is to have non-audiophiles listen! People who don't know what they are supposed to be hearing are the best guide, because the audio crowd are obsessed with mentally measuring the level of bass, precision of imaging, etc, etc.

 

I'm pleased when I can have the system running at maximum volume, and non-audio people are blissfully unaware that this is the case, and even ask for more volume; they have no trouble interacting with each other while listening - because the system is not yelling at the top of its voice, "Look at me!!"

[fas42]

31 minutes ago, sandyk said:

 

 How can they ever be truly " lifelike" while using ONLY 2 front stereo speakers , which the vast majority of Audio systems do ?

 

A competent stereo system presents lifelike reproduction because the 2 front speakers completely disappear from one's awareness - it literally becomes impossible to locate the position of the speakers just using one's ears.

 

Playback will fail the test, if there are giveaways in the sound that our ears zoom in on - it is not trivial to deceive our hearing, but can be done with sufficient attention to detail.

[fas42]

22 minutes ago, sandyk said:

 

Sorry Frank, but that depends entirely on the recording itself.

However, some multi channel recordings down sampled to Stereo may sound very close with a better than average system.

 

At a lower level of optimisation that will be the case. The goal is to advance the playback SQ to the point where the speakers always disappear, irrespective of the recording - this is something I've been playing with for many, many years, and it has always turned out that even the most unlikely recordings will 'succumb'.

 

This is how one advances the "art" - accepting a compromise simply because that's as good as one can get with the current situation, and personal knowlege, is not a useful solution.

[fas42]

7 hours ago, jabbr said:

 

My statement makes no such assumptions, nor is dependent on a particular number of speakers, particular file format, particular power supply, particular type of cable nor any other detail. I said that until we understand what is not adequately reproduced, then we cannot measure the difference between any actual system and an ideal reproduction system — the failing is not in the ability to measure but knowing what to measure.

 

What is usually inadequately reproduced is the fine detail; technically it may be all or largely there, but it is too distorted or masked by inadequacies of the playback chain - the listening brain either can't make sense of it, and then it just sounds a mess; or it is too stressful for our minds to untangle, and listener fatigue builds very quickly.

 

Objectively measuring this behaviour is not going to be easy, because it's 'buried' amongst the obvious, high level sound elements. And you can't just measure the behaviour of a low level track by itself, because the misbehaviour is due in large part from the presence of the high level content - it's a form of intermodulation distortion.

 

The listening mind has no trouble distinguishing an 'ideal', from conventional quality - all the usual adjectives and phrases immediately spring to mind; effortless, organic, natural, immersive, holographic, "you are there", "an open window", etc, etc.

[fas42]

Chopra? He was a bright boy, knew his endocrinology as well as anyone, but got buried in the spiral of modern living - "drinking black coffee by the hour and smoking at least a pack of cigarettes a day". Becoming aware of an alternative outlook, he vigorously pursued it, because he had an energetic mind ... not such a bad comparision, perhaps.

[fas42]

5 hours ago, Kal Rubinson said:

Stereo is a compromise.

 

Maybe a compromise, but can deliver remarkably impressive experiences, if the equipment used is up to it. Most people have at least once or twice come across a stereo rig "that blew their mind" - that shows what's possible, when all the usual compromises are absent.

[fas42]

32 minutes ago, bluesman said:

All reproduced music is a compromise - I've never heard a playback that was truly indistinguishable from a live performance.  I've always believed that this is because intermodulation among all notes from all instruments being played is captured in every recording as part of the source waveform.  During playback, the original "natural" IM is created again, so there's twice as much being played back as was generated in the original performance.  And it's now intermodulating anew with the source signal to generate yet more audible energy that wasn't present in the performance.  I strongly suspect that this is what throws a sonic "veil" over all reproduced music and what keeps it from sounding truly live.

 

Disagree. That "sonic veil" can be completely eliminated - because it's an artifact of the playback chain not behaving sufficiently well.

 

Typically, a rig below par will project that sonic veil; then, there's an abrupt transition in the subjective perception when the rig rises above the necessary level; whichever way the latter is achieved. Everything "snaps into focus" - this is a good phrase to use, because it's like using powerful binoculars without understanding how to stabilise on what you're pointing at, and how to fine tune the optics settings; until you get this exactly right it's an exercise in frustration, with bad blurring completely ruining the point of using a large zoom.

[fas42]

10 minutes ago, STC said:

The real problem with reproduced music is creating the position of the instruments. In live performance (unamplified) we could locate the source based on HRTF. With stereo, to recreate the positional cues of different instruments it only uses one of the HRTF cues to recreate a phantom image. This is the difference between live and playback. Real image (position) is just one spot for each source in the live performance. However, with music, the phantom image is created by two speakers. The brain is continuously locating two sources to create a single phantom image. The process is contradicting how real sound in nature behaves and therefore itself is unnatural There is no way during playback of two speakers to recreate the soundfield or soundstage will ever be natural to us unless each instruments sound is confined to a single speaker.

 

 

 

IOW, if we were to listen to live sound on the other of a wall, where there were only two openings to the other side, mimicing the positioning of stereo speakers, it would sound just like a hifi system! Of course, no-one has done this sort of experiment; because it might expose an Inconvenient Truth ...

[fas42]

The point is, what would it sound like? Because, that's largely what two microphones set up in the same position would 'see'.

 

STC feels that it would have a lack of clarity - I'm concerned with the realism of the sound; if you were to stand before a open door into a room where live music was being played, would it sound 'fake'? Now, replace that single door with two smaller doors, spread apart ... get my drift?

 

 

[fas42]

3 minutes ago, bluesman said:

Isolating the instruments greatly reduces intermodulation between & among them, and a damped environment further reduces extraneous input from resonance etc. Playing each individual instrument, closely miked for recording to minimize bleeding, through its own speaker is the best way I know of to prevent intermodulation from contaminating the recording. It's closer to having the individual instruments in the playback setting.  That's why it sounds more like real instruments, but in the playback environment rather than the recording setting.

 

Amusingly, there are some 'experimental' recordings where only the stereo mike is used, and the balance between the instruments is fine tuned simply by adjusting the distance and positioning from that mike, of each musician. Strangely enough, it sounds very realistic - must be some powerful mojo happening in the space, to stop all that nasty intermodulation contamination.

[fas42]

9 hours ago, STC said:

 

 

IMO, there is nothing wrong if the sound comes from the tweeter or woofer. That's how a speaker function anyway. In a normal two stereo setups (devoid of reflection) the image(s) will float between the two speakers. This is more natural compared to having all sound emanating from a single speaker. However, saying that sound shouldn't originate from the woofer itself is wrong as it depends on the size of the instrument itself.

 

Yes, the sound comes from the speaker. But our brain interprets all the auditory cues and positions the sound element "where it makes sense" to be. A playback system below par doesn't reproduce those cues precisely enough, with enough clarity - and the sound then obviously emerges from the speaker drivers, especially when you get a bit closer to them.

 

If one is able develop a rig to the point where the brain has enough information to continually maintain a convincing illusion, then there can be no going back. Normal stereo replay then becomes cartoon like, and has no special value as a listening experience, because it is so obviously "fake".

[fas42]

19 minutes ago, Jud said:

 

Since measurements are more accurate than the ear, I wonder what "tweaks" are needed.

 

The "tweaks" are the fairy dust holding up the bridge - because if done correctly they are actually enhancements and corrections to the poor design and implementation of the raw, audio system. An playback rig is like a bridge - the latter if not strong enough is a major failure, and may have to be torn down and replaced as a solution ... and the same methodology, for ensuring integrity of the bridge, gets the job done in audio ...

[fas42]

Some measurements are necessary, but not sufficient for good sound.

 

All relevant measurements are necessary and sufficient for good sound.

[fas42]

6 minutes ago, DuckToller said:

Frank,

which are the measurements you find necessary and sufficient for good sound???
Tom
 

 

If I knew that, I would sell it as a product ... .

 

Unfortunately, I'm relying on my ears as much as anyone here who chases optimum sound - another lifetime, perhaps I'll get around to it, .

 

As implied in mansr's response, it's detecting the influence of poor integrity of the overall system, which in part is caused by poor standards of some connections. So, there would be some sort of signal correlated noise present which needs to be measured - this would be certainly possible to detect; it just needs to be researched to come up with an ideal test signal.

[fas42]

2 hours ago, crenca said:

 

Oh, I don't believe it is useful to say that measurements are more "accurate" than the ear, or vice versa.  Different domains.  I think you can only correlate statistically.  Perhaps a psychoacoustic guy will chime in...

 

Far from "a psychoacoustic guy", but as someone who has twigged onto this term relatively late in the day - the way the brain reacts to annoying anomalies is where the real action is. My first bit of kit that performed well had "all the right measurements", which is initially why I was happy to buy it - but certainly didn't stand on its hind legs straight out of the boxes. I spent plenty of time listening to all the kit that was around back then, and I was reassured that my choices were good, even in their raw state - and then that situation progressed to a continuous cycle of learning exercises, where I would become aware of slight flaws in the sound, and tried things, to see if I could eliminate what I heard as not being quite right ... then, one day ...

[fas42]

OCD? Is that what they call an interest in debugging, these days?

[fas42]

Just noted this,

 

Quote

Fundamentally, what interests me most about amplifiers are the differences in sound created by different topologies and the characteristics of the active gain devices.

 

, from https://www.stereophile.com/content/nelson-pass-circuit-topology-and-end-science.

Immediately tells me I wouldn't be interested in his designs, because he's chasing an active tone control - just insert the right weakness, to add the required flavouring ...

[fas42]

1 hour ago, kumakuma said:

 

I was thinking more of your posting history here on this site.

 

My apologies. I didn't appreciate that this was a site not interested in various methods used by people enthused about audio for improving their SQ.

[fas42]

Dear me. How to design an amplifier ...

 

Quote

The lesson of the Xs and SIT amplifiers was that a small amount of second harmonic of a particular phase character gives a desirable sonic result. To get that effect, we altered the arrangement of the constant-current sources in the output stage to better duplicate the sound of the Xs output stage. While the .8 amplifiers still have low distortion, they do not suppress second harmonic as much as the .5 series, giving a mostly second-harmonic character at ordinary listening levels and segueing into third harmonic at higher power.

, from https://www.stereophile.com/content/pass-laboratories-xa2008-monoblock-power-amplifier.

 

So, we've reached the point in 2018 where modern electronics by a highly regarded designer are "audibly transparent" ...