Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: audibility of polarity inversion (Read 43484 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

audibility of polarity inversion

I thought it might be interesting to post this here any see if anyone has any comments...

I'm interested in the question of the audibility of polarity reversal (let's define that to mean the effect of reversing the speaker cables on each speaker, or any equivalent operation) during music playback. I've done a few little experiments with it recently, and I thought some might be interested in the results and discussion.

The first question to address is whether polarity reversal is EVER audible. Many people believe it is not, but they are mistaken. Let's see how we can make it as easy to hear as possible. First of all, it's obvious that only for an asymmetric signal (meaning one which is significantly different when inverted) can we expect to hear something. On a steady pure tone, for example, polarity can not possibly be audible because the inverted signal is identical to the original, time shifted by one-half period. So let's construct a very asymmetric signal and listen for the effect there.

To do this, take any audio program (Audacity is a good free example) and generate a pure tone at, say, 400Hz. Now generate another at 800Hz, but shift the phase by 90 degrees relative to the first. Summing those two signals gives a very asymmetric waveform. If you want to make it even more asymmetric, add another signal at 1200Hz with phase 180 degrees shifted from the first (etc.). If you look at that signal, the negative parts are of higher frequency but lower amplitude than the positive parts. The DC offset is zero. Note that it's crucial that these components be harmonic - anharmonic combinations (meaning combinations of pure tones where the frequencies are not integer multiples of some base frequency) are phase incherent (the polarity reversed signal looks the same, at least when averaged over a few cycles), and so polarity reversal can not be audible for them.

Now take that summed signal, call it test1, and make a new signal which is inverted (the invert function in Audacity does that for you). Call that test2. Now download a player with an ABX comparator (foobar2000 with the ABX plugin for example) and see if you can hear a difference.

After a little practice I found that quite easy. It sounds like one tone has a higher frequency than the other. Now let's say on a particular trial A sounds higher pitched than B. Here's where it gets really interesting - my headphones (Grado's) can be rotated so that the speakers are facing outwards, away from the ears, rather than inwards. When I wear them like that, I can still hear the difference (10/10 correct on ABX), but now B sounds higher than A!

This result has several implications. First of all, it proves that the difference I'm hearing is in my ears, not to anything in the playback chain like the computer or the phones. If the difference was due to distortion in the phones, for example, inverting them wouldn't change anything.

Second, it supports the hypothesis (based on the little I know about ears and hearing) that the reason polarity is (sometimes) audible is that the ear acts roughly as a half-wave rectifier - that is it chops off the part of the signal that is a rarefaction, say, but not the part that's a compression (it might be the opposite, I don't know - one could actually find out by determining which way the phones are moving and checking which way sounds higher pitched). The pitch change is evidently due to the fact that this signal, when rectified one way (keeping the positive part), will have higher frequency than when rectified the other way (keeping the negative part).

So polarity reversal is audible on certain asymmetric test tones, and the likely explanation is that the ear/brain acts to a first approximation as a half-wave rectifier.

Now the question of greatest interest is whether polarity is audible for recorded music. I think the answer is almost always no, at least on a system with low distortion, but I'll come to that in the next post.

Now for the issue of the audibility of polarity inversion in music.

First, I have been unable to find any peer-reviewed published evidence that demonstrates audibility for musical selections. Instead, controlled tests (e.g. Greiner and Melton) done with very high-fidelity and low-distortion (usually ESL type) loudspeakers, playing mono tracks in acoustically isolated environments, and using musical selections chosen by the researchers to have the greatest asymmetry in the waveforms (so as to make polarity as audible as possible) have failed to find audibility except for a very few special cases. Those cases were essentially individual tones played by a single acoustic intrument. If other evidence exists, please comment.

There is some anecdotal evidence that polarity can be audible in music, particularly when LPs are used as the source. One interesting possibility is that polarity is or can be audible on such systems because they themselves add a significant amount of asymmetric distortion. LPs in particular are susceptible to that - the needle moving up is not necessarily equivalent to the needle moving down, and so asymmetric waveforms can be made even more so by added distortion. Therefore the first task, if audibility can be demonstrated with music in some audio system, is to determine whether the effect comes from distortion in the playback chain, or whether it was present in the original recording. There are a number of tests that could be done to check this.

If audibility of polarity inversion can be demonstrated, the next question is whether there is any such thing as absolute polarity - in other words, is there a polarity choice which is in some sense higher fidelity or superior to the other. For the case of test tones, where polarity is demonstrably audible, there is clearly no such choice.

Now consider the case of a single acoustic source recorded with a single microphone. In that case one could control the polarity throughout recording and mastering and playback, so as to ensure that a compression wave generated by the original instrument corresponds to a compression wave generated by the speakers in the listening room. Presumably this could be regarded as the "correct" setting for polarity.

However there are two major problems that arise the moment we consider a more realistic situation. Suppose there are two mics separated by some distance which an engineer will later mix down into one track. Suppose also there is more than one source of sound (either because there are two instruments/voices, or because of reflected sound in the studio). In that case it is simply impossible to make a polarity choice that corresponds to the original polarity. The reason is that because the two mics are separated, they will be out of phase on nearly all components of the signal (recall that audio frequency sound waves have a wavelength which is of order a few inches), and will not receive the leading impulse of sound at the same time. With more than one sound source there is no way to time-align the arrival of both sources. Recall also that the phase or polarity of a sound can be inverted upon reflection off a surface, or in the case of a dipole radiator by simply moving 180 degrees around to the other side of the instrument. Therefore there is no way to time-align or phase align the two mic tracks in such a way that the mixed track preserves the original polarity in any sense, nor do audio engineers even consider doing so (as far as I know).

The second problem is with stereo - even if only two mics are used, one for each stereo track, unless the speakers are positioned exactly as the mics were in the studio, the signal that arrives at the listener's ears will suffer from phase distortions (although in this case it might still be possible to preserve at least some of the polarity information).

The only way I can see to make a recording that preserves polarity or phase is to make a binaural recording (two mics about as far apart as your ears, one for each track) played back through headphones. Maybe some sort of standard for dual-mic recording could be developed that would make it possible to arrange something like that for stereo playback through speakers, but so far as I know there are no such recordings in existence now.

Therefore I reach the tentative conclusion that polarity is not an issue to be concerened with for musical playback.

Any comments?

audibility of polarity inversion

Reply #1
I believe this issue was discussed at length recently. I didn't follow the thread very closely but I think there were some thoughts that the sound of a drum could be different with phase reversal.

audibility of polarity inversion

Reply #2
That would have been a post I made a couple years ago with a very similar-sounding name. Also I linked that post a couple weeks ago on Audio Asylum, thereby igniting a tinderbox of controversy between the always-easygoing Clark Johnson and, well, everybody else.

(EDIT: Including you, heh)

The first question to address is whether polarity reversal is EVER audible. Many people believe it is not, but they are mistaken.
IIRC, most people do not dispute that polarity inversion (aka the Wood effect) is audible for specifically designed synthetic signals. As I recall, Grenier/Melton were able to reproduce it with synthetic tones in their tests under most situations.
Quote
After a little practice I found that quite easy. It sounds like one tone has a higher frequency than the other. Now let's say on a particular trial A sounds higher pitched than B. Here's where it gets really interesting - my headphones (Grado's) can be rotated so that the speakers are facing outwards, away from the ears, rather than inwards. When I wear them like that, I can still hear the difference (10/10 correct on ABX), but now B sounds higher than A!

This result has several implications. First of all, it proves that the difference I'm hearing is in my ears, not to anything in the playback chain like the computer or the phones. If the difference was due to distortion in the phones, for example, inverting them wouldn't change anything.
I don't see how this is true. If even-order harmonic distortion existed in the phones, it could either increase or decrease the signal's assymmetry, depending on its orientation. Given the ear's sensitivity to assymmetry, your results can be completely explained by distortion in the playback chain.

This was the biggest counterargument to my original post - that I was able to get stunning ABX results, but it could only be reproduced at extremely loud volumes (which naturally makes distortion more likely), it was better on some headphones and sound cards and not others (possibly reflecting different levels of distortion), etc. Asymmetric limiting distortion, being a form of even-order harmonic distortion, can happen (perhaps quite often) for amplifiers and transducers. If one polarity was simply clipping more than another, than clearly it would be far more audible than if the clipping was equal in both polarities.
Quote
So polarity reversal is audible on certain asymmetric test tones, and the likely explanation is that the ear/brain acts to a first approximation as a half-wave rectifier.
I don't think anybody has come forward with a physical model for the audibility in the literature; your explanation is as good as anybody's. However, I strongly suspect that the behavior exists as an assymmetric soft-limiting effect rather than a rectification. That is, one pressure polarity reproduces more weakly relative to the other pressure polarity, the stronger the pressure wave is.

The acid test for any model is that if this model is inverted, on accurate playback equipment, polarity audibility should disappear entirely.
Quote
However there are two major problems that arise the moment we consider a more realistic situation.
Oh but wait! There's a third!  Reflected signals are in opposite polarity from the source signal. So once room reverberation matters, either in recording or in playback, you're already polarity-incoherent. (Another thing I learned in the earlier thread.)

I tend to agree with you that polarity doesn't really matter a whole lot. Most certainly, it cannot be used as a metric for sound quality or equipment quality, as some audiophiles commonly do. If a music selections' timbre changes notably with a polarity change, I would suspect more than anything else that it's being played back on a ratty tube amp.

In terms of actual musical effect, if one subscribes to the notion that harmonic distortion generally increases the timbre of the music, that might indicate some sort of preference of one polarity over another. That could make sense with my own observations. But at the same time, nothing's holding an artist or engineer back from deliberately inverting a signal for artistic reasons. Like you say, a mix of multiple tracks is itself a travesty for accurate soundfield reproduction.

audibility of polarity inversion

Reply #3
I've seen quite a few different pressings of the same CD where the phase has been inverted in both channels.  Whether anyone has noticed or not by listening, I don't know, but I highly doubt it!

All of these pressings have drums and few if any of these pressings feature any synthetic sounds, BTW.

audibility of polarity inversion

Reply #4
That would have been a post I made a couple years ago with a very similar-sounding name. Also I linked that post a couple weeks ago on Audio Asylum, thereby igniting a tinderbox of controversy between the always-easygoing Clark Johnson and, well, everybody else.


Great, thanks for the link - I'll take a look.

EDIT - having read that thread, I have to say I'm confused why people kept discussing the question of absolute polarity on recordings.  For example this

Quote
Phase correctness in mastering and playback is a matter of standard. Its so trivially easy to maintain, that there is no justification whatsoever to phase reversal in the recording chain. I consider this part of the matter irrelevant. Incorrect phase on a recording is simply huge label of disgrace for the recording company. Its a sign of negligent attitude. What quality can you expect from a recording company that can't even get its phase (wires) right?


is just misguided - when more than two mics are involved in making a stereo recording, what does phase reversal even mean? 

Another comment - it's not true that you can't reverse the polarity by moving around a source.  If the source is a dipole, moving 180 degrees around it exactly inverts the polarity. 

I don't see how this is true. If even-order harmonic distortion existed in the phones, it could either increase or decrease the signal's assymmetry, depending on its orientation. Given the ear's sensitivity to assymmetry, your results can be completely explained by distortion in the playback chain.


I'm not sure I follow - you say given that the ear is sensitive to asymmetry (so it's the ear), the results can be completely explained by distortion (so it's not in the ear?).  Let me explain what I had in mind, and we'll see if it makes sense.

The effect I was hearing (with the phones in the normal position) which allowed me to distinguish the two sounds was that A sounded lower than B.  If you look at the signal I was using, the positive parts of A are large amplitude and low frequency, and the negative parts are lower amplitude and higher frequency.  B is the reverse.

Now if something in the playback chain distorted, say by amplifying the positive relative to the negative, it could shift the apparent pitch of A down by increasing the amplitude of the lower frequency component.  By the same token it would shift the pitch of B up.  Of course, none of that has anthing to do with what orientation the phones were in (Grados are open-back, so there's not even an air cushion around the ear to resist the drivers).  Therefore reversing the phones shouldn't change the percieved pitch of A versus B - that can only happen if it has something to do with the ear.

So I think these results can only be explained by polarity sensitivity in the ear.  Since turning the phones around exchanges compression for rarefaction, that works as expected.

Quote
This was the biggest counterargument to my original post - that I was able to get stunning ABX results, but it could only be reproduced at extremely loud volumes (which naturally makes distortion more likely), it was better on some headphones and sound cards and not others (possibly reflecting different levels of distortion), etc. Asymmetric limiting distortion, being a form of even-order harmonic distortion, can happen (perhaps quite often) for amplifiers and transducers. If one polarity was simply clipping more than another, than clearly it would be far more audible than if the clipping was equal in both polarities.
Quote
So polarity reversal is audible on certain asymmetric test tones, and the likely explanation is that the ear/brain acts to a first approximation as a half-wave rectifier.
I don't think anybody has come forward with a physical model for the audibility in the literature; your explanation is as good as anybody's. However, I strongly suspect that the behavior exists as an assymmetric soft-limiting effect rather than a rectification. That is, one pressure polarity reproduces more weakly relative to the other pressure polarity, the stronger the pressure wave is.


I was able to do this at medium to low volumes, making distortion less likely (and see above).  Actually there are models for the ear like this in the literature - I think half-wave rectifier is more or less the standard starting point (obviously that's oversimplified, so people replace it with some smoother function that's qualitatively similar).  I can dig up a web reference if you're interested.

Good point about reflections - I did mention that, but didn't emphasize it as I was too lazy to figure out if the polarity was reversed or not.  But you're correct - reflections off a hard wall will invert the polarity.

audibility of polarity inversion

Reply #5
Quote

Phase correctness in mastering and playback is a matter of standard. Its so trivially easy to maintain, that there is no justification whatsoever to phase reversal in the recording chain. I consider this part of the matter irrelevant. Incorrect phase on a recording is simply huge label of disgrace for the recording company. Its a sign of negligent attitude. What quality can you expect from a recording company that can't even get its phase (wires) right?


is just misguided - when more than two mics are involved in making a stereo recording, what does phase reversal even mean? 
In reply to this and your earlier points on miking and polarity:

I don't really believe that there is ever such a thing as a "true" polarity that can be experienced. Even a single-miked recording exists as a sample of a soundfield that is only correct for that point, and even then it is colored by the microphone itself. When reproduced, even though the signal is mono, you'll still get an inter-channel delay between the ears that never existed in the original soundfield. This even applies when listening to headphones (where the ICD is zero but the ICD in the original soundfield certainly wasn't).

Instead, what we have are approximations to what is really going on. No matter if polarity itself is audible, there are some side effects to the reversal - such as comb filtering - that could be aggrevated.

FYI, I still disagree with the quote you mention; nobody's really proved one way or another that polarity reversal is or isn't an issue with music production. Certainly equipment polarity isn't quite standard, but also, few people have come forward with concrete examples of polarity-reversed music. It's all anecdotal or completely subjective.

Quote
Another comment - it's not true that you can't reverse the polarity by moving around a source.  If the source is a dipole, moving 180 degrees around it exactly inverts the polarity.
That's true, but very few things in life seem like "true" dipoles to me. If a sound source has the same absorbtion/reflection behavior for every two opposite directions, then you're right, but very few things in reality behave like that. Off the top of my head, the only sound sources that qualify would be:
  • Only certain kinds of speakers. They must be open-back. If multiple drivers exist, they would have to be coplanar and the drivers would have to be symmetric on that plane (ie a circle of tweeters around a single woofer). Minimal driver housing on the rear side.
  • Percussion where only one percussive surface exists, and it's flat.
Everything else - vocals, strings, etc - would have some irregular change when turned around, that won't quite be a real polarity reversal.

Quote
So I think these results can only be explained by polarity sensitivity in the ear.  Since turning the phones around exchanges compression for rarefaction, that works as expected.
Yeah, I think I might have not completely understood what you were doing. I can't find a particularly good counterexample that would cause the same behavior you're seeing but exist entirely in the playback-distortion domain. So I can no longer dispute this method (for now  ). Very clever too.

Quote
I was able to do this at medium to low volumes, making distortion less likely (and see above).  Actually there are models for the ear like this in the literature - I think half-wave rectifier is more or less the standard starting point (obviously that's oversimplified, so people replace it with some smoother function that's qualitatively similar).  I can dig up a web reference if you're interested.
Sure.

audibility of polarity inversion

Reply #6
There is some justification for arguing that mechanisms within the ear exist, because it seems clear that positive going displacement of the basilar membrane selectively fires more inner hair cells.

But it's not that simple, of course, because that measurement is after the cochlear filtering occurs, and so you need something assymetric in a narrow bandwidth.

On top of that, acoustics makes it hard to determine what's going on.  Clearly, things like drums are heavily assymetric on attacks, and routinely put out over 140dB peak SPL at the drum. This means that in fact the air is quite nonlinear at that level.  In fact, if you listen to a set of drumbeats and move back, you will notice a variety of timbre effects, some of which are, I believe, due to intermod due to air nonlinearity. Of course a great deal of effect comes about due to capturing reverberation and early arrivals as well as sources from more parts of the drum (drums are nothing like omnidirectional).

On top of that, most speakers are rather (I'm being polite and understated there) assymetric, which means that even a signal that might otherwise sound the same won't because of speaker nonlinearities.

On the transducer nonlinearity front, quite a few tests of extraordinary hearing claims have foundered, in particular diaphram bending, etc, resulting in subharmonics that you can hear, or intermod that you can hear.

This is a bleeping hard test to run well.
-----
J. D. (jj) Johnston

audibility of polarity inversion

Reply #7
Quote
That's true, but very few things in life seem like "true" dipoles to me. If a sound source has the same absorbtion/reflection behavior for every two opposite directions, then you're right, but very few things in reality behave like that. Off the top of my head, the only sound sources that qualify would be:
  • Only certain kinds of speakers. They must be open-back. If multiple drivers exist, they would have to be coplanar and the drivers would have to be symmetric on that plane (ie a circle of tweeters around a single woofer). Minimal driver housing on the rear side.
  • Percussion where only one percussive surface exists, and it's flat.
Everything else - vocals, strings, etc - would have some irregular change when turned around, that won't quite be a real polarity reversal.


Yeah, I agree - there aren't too many perfect dipole radiators around.  However there also aren't many pure monopoles, and something like a plcuked guitar string is probably better approximated as a dipole than anything else.  In any case, the point is that polarity is at least partly a function of relative position, further undermining the idea that there's such a thing as the natural or correct choice.

Quote
So I think these results can only be explained by polarity sensitivity in the ear.  Since turning the phones around exchanges compression for rarefaction, that works as expected.Yeah, I think I might have not completely understood what you were doing. I can't find a particularly good counterexample that would cause the same behavior you're seeing but exist entirely in the playback-distortion domain. So I can no longer dispute this method (for now  ). Very clever too.


Thanks!

I can't seem to find the pdf document I looked at before that had a nice discussion of mathematical models for the reaction of the ear to pressure, but here's an interesting link:

http://epl.meei.harvard.edu/~keh/cd846/

Have a look at page 7 of the fourth lecture (Inner Ear II: Physiology) and look at the upper plot for the IHC (which stands for inner hair cell, I think).  Not a bad approximation to a half-wave rectifier...

audibility of polarity inversion

Reply #8
Hey, it's awesome that someone was able to do such a scientific test to prove that polarity inversion is audible (the 400 Hz + out-of-phase 800 Hz pseudo-stationary test). Thanks, opaqueice.

I don't have proof of anything (no rigorous ABX) but from years ago I have thought absolute polarity was audible. I think it's audible on many contemporary songs. Because I believe that absolute polarity is audible on any high-freq transient attacks. For example, a plucked guitar string, a triangle chime, etc. I learned to hear (or imagine) this on real music, and by physically switching polarity to my speakers by reversing the wiring.

I also tried to test this out on kickdrums..."low freq" transients. But I was unable to hear (or imagine) any difference.

Because of this test, I wire all my 2-way speakers in REVERSE polarity (they are in phase with each other but reversed from the way they are supposed to be).

In case you didn't know, almost all loudspeakers wire alternating drivers in reverse phase to each other. It is a basic principle of speaker design. A 3-way speaker always has the woofers in correct phase, the midrange in reversed phase, and the tweeter in correct phase. I therefore wire my 3-way speakers the normal way.

Most 2-way speakers (at least the ones that I tested) wire the woofers in correct phase and the tweeters in reversed phase. Based on my personal (unproven) listening tests, absolute polarity matters way more for the highs so I purposefully reverse-wire all my 2-way speakers.

Another good reason to do this is that if you play your 2-way and 3-way speakers at the same time (if you had a mismatched surround system, for example, or you just wanted to play multiple speakers at once which I do sometimes for fun) you MUST reverse wire one of them or the final sound will be extremely weird. This is how I first noticed polarity problems, and this is how I checked the polarity of my 2-way and 3-way speakers (by playing them simultaneously and listening for cancelations in certain freq ranges). Note that this is a lot easier than listening for absolute polarity, so anyone can do it and no one should doubt that it works. If you reverse-wire a 2-way and 3-way speakers and play simultaneously, the deep bass will disappear but at least the sound is fairly normal otherwise. If you wire them both the same, the bass will be powerful but all other sounds will sound really wacked out (partially canceling each other).

Oh, I forgot to mention what the effects of polarity inversion of high-freq sharp transient attacks/notes sounds like to me. Again, I could be imagining everything, but I feel that when wired with the correct polarity (note that this means REVERSED polarity for 2-way speakers) they sound sharper, crisper, and I think I can often hear slightly more detail afterwards as well (but after a while the note will become tonal and of course, at that time absolute phase can no longer be heard). In the other polarity, the high-freq transients sound more muddled to me, slightly.

Also, I badly damaged my hearing since then (test was several years ago) so I have no idea if I can still hear the same thing or not. Oh well. I will continue to hook up my speakers in the same way regardless.

 

audibility of polarity inversion

Reply #9
Porcupine,

For all your "I think I hear this but didn't ABX" posts, there must be more friendly forums for you! That's not a criticism - I just wonder why you enjoy the inevitable pain that comes from constantly ignoring this forum's founding principle?!

As for speaker drivers out of phase - if the signal, as delivered to the cones, was out of phase between tweeter and mid, then that would cause horrible destructive interference in the cross over region. This really can't be the case, surely? I'm speaking outside of my knowledge - I did some of the theory of analogue cross overs a long long time ago, but the only ones I ever worked on were digital, so of course all drivers were very much in phase, and sometimes phase corrected.

Cheers,
David.

audibility of polarity inversion

Reply #10
Porcupine,

For all your "I think I hear this but didn't ABX" posts, there must be more friendly forums for you! That's not a criticism - I just wonder why you enjoy the inevitable pain that comes from constantly ignoring this forum's founding principle?!

As for speaker drivers out of phase - if the signal, as delivered to the cones, was out of phase between tweeter and mid, then that would cause horrible destructive interference in the cross over region. This really can't be the case, surely? I'm speaking outside of my knowledge - I did some of the theory of analogue cross overs a long long time ago, but the only ones I ever worked on were digital, so of course all drivers were very much in phase, and sometimes phase corrected.

Cheers,
David.

Since the most used filters for speakers are still 12db/octave this second order filter requieres changing of the polarity in a typical 2-way design. All speaker kits i know do this by changing the polarity of the tweeter.
More-way systems often are different in polarity between the chassis cause of mechanical behaviour of the phisycs of the chassis and mounting.
But for a typical 2-way system i second, that most of the times the tweeter has to be changed in polarity.
Is troll-adiposity coming from feederism?
With 24bit music you can listen to silence much louder!

audibility of polarity inversion

Reply #11
To amplify that a little, the point is that a 2nd order crossover inverts the polarity of the signal - or more precisely it advances the phase by 180 degrees.  Inverting the polarity of one driver will then restore phase coherence for steady tones.  However it will not restore it for transients, because there is a time-delay.  My guess is it's better to wire the woofer with the standard polarity so as to avoid confusion when a sub-woofer is added, but maybe there's another reason?

A 4th order crossover will impart a 360 degree phase change, so the drivers can be wired with the same polarity, but again there is a time delay (even longer this time).  Even a speaker without crossovers will mess up the time-domain signal to some extent because there are additional time-delays and phase shifts due to the mechanical damping of the drivers! 

On another topic, I want to correct something I mentioned in my first post, and that Axon had said as well.  It is NOT true that the polarity of sound waves reflecting off hard surfaces inverts.  Instead, the polarity stays the same for perfect reflections.  However for reflections off of openings (for example if you have an open doorway in your listening room) the polarity DOES invert.

audibility of polarity inversion

Reply #12
Yeah, I'm making the mea culpa rounds on that one. Thanks opaqueice, and RickM on AA. I'll be updating the original thread shortly to make sure nobody else is misled.

Most informative Audio Asylum post EVAR! (and its replies)

http://www.audioasylum.com/forums/critics/...es/3/32302.html

audibility of polarity inversion

Reply #13
Eh? No, the polarity DOES invert when a sound wave is reflected off a hard surface. You were right the first time, if I'm not mistaken. The sum of a traveling wave in the +v direction plus the sum of an equal amplitude, but negative in sign, wave traveling in the -v direction, which results in a standing resonance wave. Well, you could define the mathematical functions in a number of ways and say that the reflected wave is the same polarity as well, I suppose.

But for the case of a transient pulse wave hitting a hard surface, it does reflect back with the negative amplitude, that's an absolute fact I think and not subject to the way you write your math. So you were right the first time. And a transient pulse hitting an "open" surface reflects back with the same sign amplitude.

But either way, I don't see how that matters to the main issue. Reflections are reflections, they have a time-delay and your ear can interpret them as echo effects. The polarity and phase is not an issue as the length the wave travels to hit the wall and bounce back to your head makes all that irrelevant, as by then the "position" of the 1st/2nd/etc reflection sine waves are unrelated to the original. They don't necessarily hamper your ability to hear absolute phase at all. Reflected waves are also softer than the original wave, often hugely so, depending on the positioning of the speakers in the room, your position, the material the wall is made out of, etc.

2Bdecided, what you say does have a point. Well, I *have* been talking on this forum much less as of late. I have a lot of posts but most of them were when I just appeared here.

audibility of polarity inversion

Reply #14
Did you even read the links I posted?

The explanation that convinced me was: if polarity did change in a rigid surface reflection, conservation of mass would be violated. A positive wave pulse indicates more mass. A negative wave pulse indicates less mass. If polarity did change, then some air molecules are vanishing without a trace. Clearly that can't happen.

It matters because reflection inversion is used as an argument to assert that after 1+ room reflections, the polarity of a sound source is completely incoherent, and that one reason why polarity reversal is so hard to detect speakers is because of those reflections. Well, the reflections are still going to cause some decoherence, but certainly not as much as before.

audibility of polarity inversion

Reply #15
I didn't read your links, nor do I really feel like I want to. Also, the weird explanation you just gave sounds ridiculous to me. Sure, mass is conserved, but traveling waves have little to do with whether mass is conserved or not in the medium that carries them. I think in high school, most people do this experiment in physics class where two people hold a rope, and someone pushes a pulse to the other guy and watches it bounce back with negative amplitude. Sorry, I am not trying to be insulting or arrogant, in any way. If you don't believe me, that is fine.

I stand by what I said in my previous post, although if it's incorrect, so be it.

audibility of polarity inversion

Reply #16
The complete explanation of phase changes when sound wave reflects:

Quote
The phase of the reflected sound waves from hard surfaces and the reflection of string waves from their ends determines whether the interference of the reflected and incident waves will be constructive or destructive. For string waves at the ends of strings there is a reversal of phase and it plays an important role in producing resonance in strings. Since the reflected wave and the incident wave add to each other while moving in opposite directions, the appearance of propagation is lost and the resulting vibration is called a standing wave.

When sound waves in air (pressure waves) encounter a hard surface, there is no phase change upon reflection. That is, when the high pressure part of a sound wave hits the wall, it will be reflected as a high pressure, not a reversed phase which would be a low pressure. Keep in mind that when we talk about the pressure associated with a sound wave, a positive or "high" pressure is one that is above the ambient atmospheric pressure and a negative or "low" pressure is just one that is below atmospheric pressure. A wall is described as having a higher "acoustic impedance" than the air, and when a wave encounters a medium of higher acoustic impedance there is no phase change upon reflection.

On the other hand, if a sound wave in a solid strikes an air boundary, the pressure wave which reflects back into the solid from the air boundary will experience a phase reversal - a high-pressure part reflecting as a low-pressure region. That is, reflections off a lower impedance medium will be reversed in phase.

Porcupine, you have to reconsider your stand on this... It's the scientifical fact. Acoustic impedance of medium that wave strikes does matter.
If age or weaknes doe prohibyte bloudletting you must use boxing

audibility of polarity inversion

Reply #17
Oh. Yeah, that makes sense. Now that it's posted clearly here, I actually remember learning about that before, too. But I had forgotten...that the physics for longitudinal air waves is slightly different than for transverse string waves.

You're totally right, eevan. Thanks for clearing it up.

-----

Axon, by the way though, either way it doesn't have much bearing on whether or not the reflected waves hamper the listener's ability to detect absolute phase of the sound. Because like I said before, the reflected waves have to travel some extra distance and there is a time delay and phase delay before they arrive at the listener's position. The reflected waves are also softer in amplitude (either because the wall absorbed some of the energy, or simply due to traveling farther and losing amplitude inversely proportional to distance^2, or something like that). The reflected waves are just echo/reverberations. Even if there were open doorways that created reversed-polarity reflections, your ear can still hear the absolute phase of the direct initial wave.

audibility of polarity inversion

Reply #18
Even if there were open doorways that created reversed-polarity reflections, your ear can still hear the absolute phase of the direct initial wave.


I think a bigger problem than reflections is the mixing of several tracks down into one.  Nearly all recordings are made that way, and when you have more than one mic and more than one sound source it's impossible to time-align the tracks so that the polarity is preserved.  At best on a given track you could preserve polarity for one instrument, but even that is questionable because the polarity of an instrument depends on listener/mic location (a guitar waveform will look different when the string is plucked towards rather than away from the listener).

Only with recordings made with a single mic per channel, or perhaps with one ambient mic and several monitor mics, could some semblance of polarity be preserved.

audibility of polarity inversion

Reply #19
Only with recordings made with a single mic per channel...could some semblance of polarity be preserved.
I'm not an expert on how professionals master their music tracks, but I had thought that this was how most contemporary music is made these days anyway...one mic per instrument, either recorded in a studio, or synthesized electronically from start-to-finish...then mixed together afterwards "electronically".

Even in the case of a live recording with multiple microphones, although absolute polarity will be polluted a little like you say, I still think absolute polarity comes through somewhat. For example, one instrument which created a sharp transient pulse in the positive direction, would get recorded by each mic as a transient pulse in the positive direction (approximately...with perhaps some slight change as you say, but the mics will probably be at least on "one side" of the instruments, so the polarity from one mic to the next shouldn't be a complete negative reversal). Added together the pulse might be a bit smeared but should still consist of pulses in the positive direction, it just won't sound as sharp and might consist of several, ultra tiny time-lagged pulses one after another, superposed.

audibility of polarity inversion

Reply #20
Just my $0.02


I thought it was pretty much a truism among engineers that you can hear phase cancellation.

I know when I am using a console with a mono button I can certainly hear cancellation. (often just moving the mic a little bit will fix this.) Cancellation tends to be most noticeable in the 1k and above range so if you think about the size/length of the waveforms in question you are talking about millimeters in difference. I have had very few instances where I needed to use phase flip for this. (Only those times where I didn't hear the cancellation till the recording was halfway done) Those who are questioning the validity of it's EXISTENCE should keep in mind that cancellation is most apparent in instruments of similar frequency.

In a multitrack recording there is always going to be cancellation. Of course. Sometimes it's even a blessing. But when my hat/snare/ride are canceling each other on the overheads and ruining my stereo image, that's when I need to worry about it. When your mix comes out sounding all lop-sided due to cancellation it's no fun.


One last thing. Make sure you don't confuse "polarity inversion" with "phase." One is electronic while the other is based on "time shift." Although polarity inversion (phase flip) can correct the SYMPTOMS of phase shift, it does it by re-polorizing the signal, not by shifting it's timeline.

audibility of polarity inversion

Reply #21
Hmm, I just have one questions with regards to this - on a speaker (like from an old stereo system) is the wire with the marking + or - ? One is plain and one has a black line down it. There doesn't seem to be a way to open these to see if there are any markings on the speaker, so I'm hoping there's an industry standard on this sort of thing.

audibility of polarity inversion

Reply #22
1. There are markings on every speaker.
2. The wire with the black stripe should be the "+" one.
If age or weaknes doe prohibyte bloudletting you must use boxing

audibility of polarity inversion

Reply #23
One last thing. Make sure you don't confuse "polarity inversion" with "phase." One is electronic while the other is based on "time shift." Although polarity inversion (phase flip) can correct the SYMPTOMS of phase shift, it does it by re-polorizing the signal, not by shifting it's timeline.


Part of the problem here is no one knows what makes polarity audible  - is it the transient at the start of a note, is it asymmetry during the steady-state part of a tone, is it harmonic distortion in the playback system, is it audible at all on real music?  If it's the transient at the beginning of a note and you want to preserve that, then flipping the polarity switch on any mic is bad.  Instead if you have a problem with phase cancellations you should either move the mic or time-shift the recording by 1/2 a period.  If on the other hand it's a steady tone issue, then flipping the switch is fine. 

Hmm, I just have one questions with regards to this - on a speaker (like from an old stereo system) is the wire with the marking + or - ? One is plain and one has a black line down it. There doesn't seem to be a way to open these to see if there are any markings on the speaker, so I'm hoping there's an industry standard on this sort of thing.


I don't think there's any industry standard even on what red or + means.  As was mentioned above speakers often invert polarity on one of the drivers in order to prevent phase cancellations.  Just make sure you hook up both speakers in a consistent way (so they're in phase with each other), and if you want to know more you'll probably have to open up the speakers and look at how the crossovers are wired.

audibility of polarity inversion

Reply #24
Quote
Part of the problem here is no one knows what makes polarity audible - is it the transient at the start of a note, is it asymmetry during the steady-state part of a tone, is it harmonic distortion in the playback system, is it audible at all on real music? If it's the transient at the beginning of a note and you want to preserve that, then flipping the polarity switch on any mic is bad. Instead if you have a problem with phase cancellations you should either move the mic or time-shift the recording by 1/2 a period. If on the other hand it's a steady tone issue, then flipping the switch is fine.




I understand what you mean. I think the answer here is really all of the above. The real question is: At what point does it become an issue worth worrying about?

is it the transient at the start of a note?
It can be.

is it asymmetry during the steady-state part of a tone?
It also can be. (This one is usually the most audible)

is it harmonic distortion in the playback system?
This one is a mostly definite possibility, given the tendency of filter circuits to introduce phase shift. Use of phase linear EQ in mastering is essential in my workflow.

Better probably to concentrate on what we DO know. We do know that inverse sine waves cancel.

We also know that inverse complex waves when summed together warp each other's shape to form a new one. (the side effect being cancellation of some frequencies)

So, it follows that in a multi-layered multi-track recording there will be lots of different waves competing. (pushing each other up and down, to and fro to form a more complex waveform) Common sense dictates that these waveforms are going to cancel somewhere. (Even if everything *IS* in phase) Just from the sheer mass of tracks piled on top of each other.

For this problem, we have "mastering equalizers." The good news is, unless you are recording some odd 80's  German industrial band that performs in all square, sine and triangle waves... cancellation never gets rid of the sound entirely. So, we can just go in and boost/cut frequencies until we get most of our desired frequencies back. Of course, it's best to have this at the outset with honest to god phase coherency but we as mere mortals do make mistakes so... it follows that we have to eventually correct them. Otherwise mastering engineers would all be out of a job.

I just realized you may be taking a different angle on this than I originally thought. If you mean what makes CANCELLATION audible? Why phase inversion CANCELS AT ALL?

It's because sound waves are consistent of compression and rarefaction of air. (In a speaker system this means it's either PUSHING or PULLING.) But it can't do both at once, so it stays still. Polarity Inversion itself is only audible because of the physical limitations of the speakers to be able to mechanically move dynamically back and forth. Kind of like, if you drop a pebble in water it makes ripples... but if you were able to drop a hula hoop shaped rock AROUND the other rock at the same time as the first rock it would keep the other rock's ripples from rippling. (What a weird metaphor)

You could also maybe think of it like this. When you wire speakers out of phase and have them face each other, Imagine they are water pumps and one is pumping water out while the other is sucking it in. Even if you point the two water pumps towards yourself, some of the water from one is still going to pump into the other. (Air is a liquid after all, albeit a fast moving one)

BTW, this trick is used quite often for singers that can't sing with headphones. (wiring the speakers out of phase and having them face each other.) Try it some time. Nothing, or next to nothing will bleed from the speakers into the singer's mike. Anything that DOES make it into the mike will only be from secondary and tertiary reflections. (Because they are phase inverted) Yet, with foam in the right places easily fixed.

I feel like I'm rambling here. I am half-awake. (Am I completely off base here as to the question?)