How audible is distortion?

[Wombat]

These cardioid type speakers are designed to meet a specific goal (uniform dispersion at all frequencies), so they are not, IMO, fooling us in that regard: I think they are just meeting that goal better than other speakers.

The one area where psychoacoustics might come into operation is the bass limiting at very high SPLs, where the system may do things dynamically, and maybe there are specific tricks necessary to make it more transparent. Then yes, I would see that a smaller box is being pushed into an area where it is struggling - but by all accounts this is only at very high SPLs indeed. The Kii uses motion feedback, so its bass is probably counter-intuitively strong and clean anyway.

Motional Feedback(Philips) requires sophisticated amplification, power and bass driver design. Still, a small box and if you cram it all(the amps and power supplies) into the small box it is more problematic. Limiting is distortion, plain and simple. Psychoacoustics is trickery, is it not?

 

[Cosmik]

Motional Feedback(Philips) requires sophisticated amplification, power and bass driver design.

I would say: it could be bog standard amplification, some cleverness regarding electronics, sensing, feedback and stability with the aim of making the cone do exactly what it's supposed to do regardless of the effects of the box and signal. No psychoacoustics though.

Still, a small box and if you cram it all(the amps and power supplies) into the small box it is more problematic. Limiting is distortion, plain and simple.

Yes, but if it only kicks in at high SPLs and the alternative is hard distortion then I don't see a problem. The box is only too small when it affects the sound and in this case it maybe has no effect until very high SPLs and/or continuous use. Maybe it would be completely honest to add a limiting indicator lamp, but then again as the listener do you really want to see it? And at what point should it turn on? At 0.0001% 'limiting effect', 0.1%, or 10%?

Psychoacoustics is trickery, is it not?

I would say that psychoacoustics is what goes on in our heads. Manufacturers may like to talk about their use of applied psychoacoustics, but I suspect many of them are just talking the talk, or maybe retrospectively fitting an explanation to some quirky aspect of their product. I think it's a lot simpler than is commonly thought.

 

[Wombat]

I would say: it could be bog standard amplification, some cleverness regarding electronics, sensing, feedback and stability with the aim of making the cone do exactly what it's supposed to do regardless of the effects of the box and signal. No psychoacoustics though.

Yes, but if it only kicks in at high SPLs and the alternative is hard distortion then I don't see a problem. The box is only too small when it affects the sound and in this case it maybe has no effect until very high SPLs and/or continuous use. Maybe it would be completely honest to add a limiting indicator lamp, but then again as the listener do you really want to see it? And at what point should it turn on? At 0.0001% 'limiting effect', 0.1%, or 10%?

I would say that psychoacoustics is what goes on in our heads. Manufacturers may like to talk about their use of applied psychoacoustics, but I suspect many of them are just talking the talk, or maybe retrospectively fitting an explanation to some quirky aspect of their product. I think it's a lot simpler than is commonly thought.

I think it is more complex than we are lead to believe. One size doesn't fit all.

 

[Cosmik]

I think it is more complex than we are lead to believe. One size doesn't fit all.

*Everyone* here acknowledges that sighted testing is just anecdotal. Interesting, but anecdotal. And we all accept that we can 'hear things' depending on our mood, expectations etc.

But at the same time, all of the things that listening test-oriented people spend their time investigating (without any conclusive results at all as far as I can tell) are indulgences of audiophile hobbyists and their ravings and superstitions. It is this made-up stuff that makes people think it is complex; that there is 'something' mysterious going on, where lower distortion doesn't automatically mean it sounds better or it is necessary to sample audio at 384 kHz, etc.

Listening tests will never demonstrate it either way, because of the variability and 'mental noise' of the listeners. Only pure reason can make any headway with it.

 

[Fitzcaraldo215]

Also, the microphones dotted in and around the orchestra are not 'hearing' what audience individuals hear.

Agreed, the closest being binaural, dummy head pickups, subject to differences between its fixed HRTF and the infininite variations of it among humans.

Yes, no single mic in space hears as humans do. But, what if a spatial array of mics is set up to pick up the sound field in the hall referenced to a single point, rather than trying to hear like humans? Then, if a similarly configured speaker array is used on playback, and barring major distortions by the room and speakers, ideally we will have something that will get close to reproducing at our ears the actual sound field heard in the hall, consisting of both direct and reflected sound. This is the principle behind discretely recorded Mch.

But, yes, engineers often attempt to augment this simple principle with added mics, including spot mics in the orchestra. Judiciously applied, these mixed in additional channels may or may not be of some benefit in terms of tonality and added detail. However, adding too much from those added feeds will distort the spatial perspective of the reproduced sound field.

I do have some 5-mic ITU-array recordings produced non-commercially in parallel with recording sessions of Channel Classics and other labels by a Grammy-winning engineer friend in DSD256. Others have heard them, including JA at Stereophile and friends including recording reviewers, Kal and Andy Quint. Our conclusion: simply the best we have heard to date in reproducing a sense of live performance.

 

[Blumlein 88]

Talk to J_J about perceptual soundfield reconstruction.

 

[RayDunzl]

I'm glad we are all agreed that Microphones don't hear.

A likely corollary would be that Ears don't measure.

I can, however, appreciate the synergistic relationship they have going on here at my place, though.

 

[Fitzcaraldo215]

Talk to J_J about perceptual soundfield reconstruction.

If he has something to add here, that would be great.

But, what I have described above is the simple, basic model used for decades now commercially in movies and Mch music recordings. There is nothing radical or extremely high tech about it, other than that it throws stereo-centric audiophiles into extreme agita and disbelief. I have thousands of music and video recordings employing the concept, some with the sometimes questionable embellishment of added mic channels in the mix, but subjectively not too many bad ones. And, on playback all it takes as a minimum is a cheap AVR and 5 or more speakers replicating the angular channel layout of ITU for those recordings.

Even better potential reproduction of the sound field in the hall is theoretically possible with "Immersive 3D" codecs, extending the xy space vertically into z, or height space. My experience so far with that is there are sharply diminishing returns, as would seem logical, given the geometry of our binaural hearing. We have empirically more aural acuity, localization, etc. in the xy space and much less in the height dimension with only two ears and a brain. So, I am not waiting around for Immersive 3D to establish itself as a commercially viable format, one which is demonstrably here to stay with a critical mass of convincing recordings. For me, even simple xy, 5/7.1 Mch offers significant advantages over stereo, which is just one dimensional x, width, with added phantom depth behind the plane of the speakers. But, Mch has that, too, as a starting point, plus a whole, 'nother added dimension.

Sorry to keep harping on this Mch stuff, guys. The topic is the audibility of distortion. While traditional distortion as frequently measured is THD, IM, etc., and those traditional measurements are vitally important, but it seems to be increasingly difficult, though not impossible, to find components that grossly violate the norms of accepted audibility. The emphasis of those traditional measurements is the purity of a mono channel, or two to see if they match adequately. Even if the problems of traditional distortion are not quite totally solved, there remains the question of why doesn't the very best stereo in the home, awesome and low distortion though it might be, sound more like live?

My answer to that is the spatial distortion produced by only 2 channels, which simply discards, truncates or distorts via redirection much useful information in the sound field we naturally hear live, pure and simple.

 

[Blumlein 88]

If he has something to add here, that would be great.

But, what I have described above is the simple, basic model used for decades now commercially in movies and Mch music recordings. There is nothing radical or extremely high tech about it, other than that it throws stereo-centric audiophiles into extreme agita and disbelief. I have thousands of music and video recordings employing the concept, some with the sometimes questionable embellishment of added mic channels in the mix, but subjectively not too many bad ones. And, on playback all it takes as a minimum is a cheap AVR and 5 or more speakers replicating the angular channel layout of ITU for those recordings.

Even better potential reproduction of the sound field in the hall is theoretically possible with "Immersive 3D" codecs, extending the xy space vertically into z, or height space. My experience so far with that is there are sharply diminishing returns, as would seem logical, given the geometry of our binaural hearing. We have empirically more aural acuity, localization, etc. in the xy space and much less in the height dimension with only two ears and a brain. So, I am not waiting around for Immersive 3D to establish itself as a commercially viable format, one which is demonstrably here to stay with a critical mass of convincing recordings. For me, even simple xy, 5/7.1 Mch offers significant advantages over stereo, which is just one dimensional x, width, with added phantom depth behind the plane of the speakers. But, Mch has that, too, as a starting point, plus a whole, 'nother added dimension.

Sorry to keep harping on this Mch stuff, guys. The topic is the audibility of distortion. While traditional distortion as frequently measured is THD, IM, etc., and those traditional measurements are vitally important, but it seems to be increasingly difficult, though not impossible, to find components that grossly violate the norms of accepted audibility. The emphasis of those traditional measurements is the purity of a mono channel, or two to see if they match adequately. Even if the problems of traditional distortion are not quite totally solved, there remains the question of why doesn't the very best stereo in the home, awesome and low distortion though it might be, sound more like live?

My answer to that is the spatial distortion produced by only 2 channels, which simply discards, truncates or distorts via redirection much useful information in the sound field we naturally hear live, pure and simple.

I don't know how PSFR would compare to object oriented Immersive 3D codecs. @j_j might discuss it as he has in the past here at ASR.

It was not terribly difficult to do. 7 channels (two were up and down) folded into 5 channels for playback. Very simple math performed on those channels. It would be a multi-channel version of stereo with a pair of microphones. Only one made with reference to our perceptual processes which would work better than what Dolby and the main competitor DTS while being straightforward and simple instead of always changing formats.

 

[j_j]

I don't know how PSFR would compare to object oriented Immersive 3D codecs. @j_j might discuss it as he has in the past here at ASR.

It was not terribly difficult to do. 7 channels (two were up and down) folded into 5 channels for playback. Very simple math performed on those channels. It would be a multi-channel version of stereo with a pair of microphones. Only one made with reference to our perceptual processes which would work better than what Dolby and the main competitor DTS while being straightforward and simple instead of always changing formats.

Well, seeing as I'm neck-deep in this stuff right now, I'm not going to comment. Sorry, but you must know that drill by now.

 

[peng]

Play a trumpet tone.

Measure it with an analyzer that registers "harmonic distortion".

Big numbers will display.

The closest thing I've found here at the ranch to a "distortionless instrument" (as measured in this manner) is a carefully blown beer bottle. A nearly pure sine wave was the output.

Interesting point, but I would not consider such natural harmonic products "distortions", though the analyzer would naturally not tell the difference, unless programmed to recognize each musical instrument's (unamplified/processed types) harmonics/overtones. Imo, there is a difference between harmonics and harmonic distortions.

 

[j_j]

Interesting point, but I would not consider such natural harmonic products "distortions", though the analyzer would naturally not tell the difference, unless programmed to recognize each musical instrument's (unamplified/processed types) harmonics/overtones. Imo, there is a difference between harmonics and harmonic distortions.

An old-timey "distortion analyzer" simply nulls out the highest single tone in a signal, that's all. So for a trumpet, it's a completely useless device.

This is why one uses spectrum analysis of "before" and "after".