Aloha,

An interesting question came to my mind the other day, after putting together informations from different sources. First of all, let me write down the statement, that intiated this whole topic. I could quote the source, but it would take some time, so if you are really interested where did I get it, ask me, and I will look it up - again.
Statement:
"With a very good speaker & amplifier combination you can literally see the instruments in front of you in their exact position."

Now there are some assumptions, that might be true or not, because this is only how I understand things - they are the combination of certain university subjects dealing with "signal processing", some topics here or there on the Internet and my own assumptions. Please correct the mistakes, if any.
Assumptions
A1: The position of a single instruments (or audio source) is determined by our brains based on the phase difference between the two signals that our ears percieve
A2: This phase difference changes with every bounce or reflection or at the borders between two different mediums
A3: Until the audio reaches our "perceiving mechanism", i.e. the eardrum (tympanic membrane), it bounces (and attenuates) several times on surfaces, including on the bumpy surface of our ears
A4: This surface is unique, you won't find two persons having ears with exactly the same curves and bumps
A5: The scientific parameters of the whole audio processing chain (analogue source -> recording -> ... -> ... -> medium -> player -> amplifier -> cables -> speaker -> environment -> listener) is very important, if we need to retain all the details of the audio information
A6: The first part of the chain (up to, but not included) the medium is close to perfect (or at least should be)
A7: The consumer can't really influence this first part of the chain

And now my own conclusions. As before, please correct the mistakes and also describe what was wrong: the conclusion itself, or the supporting assumptions.
Conclusions
C1: Great care must be taken when choosing the components for the second part, but the scientific parameters (distortion, linearity) are of big help, and should be of primary concern (next to the cost)
C2: Since your own environment can't be "measured" or rather it is very difficult and expensive, a listening test in your own environment is something that can turn your decisions upside down. Still, it won't make a better equipment (in terms of parameters) sound worse, but the perceived difference won't be as big as in the numbers (BIG QUESTION MARK HERE)?

And now the tough part, the still open questions:
Q1. How do you measure the phase difference between the signals in a music? Is there a way to tell if the recording is completely in phase with the original?
Q2: In theory, are two mics and two speaker enough to provide the "ultimate" listening experience, provided that the "band" is in front of you? Would you be able to tell the position of every single instrument?
Q3. Does a bit-perfect output mean that also the phase is being kept? If this is yes, the lossless compressions keep the phase, i.e. they keep the instrument position.
Q4: Do also lossy compressions keep the phase and position? When we say, that the compression is transparent, does that only mean that "in the current environment and processing chain" the recording can't be distinguished from the original source. But, say, have I had much higher quality eq (and ears), I would be able to distiguish the compressed and uncompressed music, if for nothing else, than because it would be "less blurred" and the position of the instruments were more exact/clear?

So, audio geeks and professionals, please let your opinion be heard!

I think you've missed out one of the most important aspects here. When music is being recorded, the instruments are usually recorded individually and then the sound engineer mixes it together to sound good. It is not mixed with the intention of preserving the direction the sound is coming from.

However, I can see where you're coming from. It's something like "is there a way of recording a band playing live, such that when it is played back, the sound is reproduced in such a way as to preserve all of the subtleties and phase differences that enable us to mentally position the players?".

My gut feeling says that this would be totally impossible to implement in practice, however, I don't know enough in detail about how the human ear physically places sound sources so I would like to hear an expert's opinion on the subject.

Sam

Moderation: Removed excessive quotation. Keep it reasonable.

Yes, you're right, with studio recordings this is practically impossible, and it was good to point this out.
And yes, your formulation of the question is what I am looking for.

This gives us a new conclusion:
C3: It means that practically there will never be a perfect recording. Either because the quality is high, but the phase is lost (studio recording of separate instruments), or the phase/position is retained, but the quality is low (live recording).

From this perspective my question is pointless, but I would still like to know whether my assumptions and conclusions were right.

PS: This topic should've gone to the Scientific R&D section. Admins, please move it.

Moderation: Moved to Scientific/R&D Discussion.

Have you read about head related transfer functions and binaural recordings? These subjects are good background for your question.
Matt

Check this page about sound reproduction from Siegfried Linkwitz

Good link, thank you for posting it. It is worth to note his words: "the effect of being there is easily achievable", but he also complains of unfilterable noise in such a setup. C3 is correct and now if only we could work out a technology that would not pick up unnecessary noise or simulate all the changes different music sources go through until they reach your ears.

I think you're putting too much emphasis on phase. There is far more information than phase which our mind uses to assign position to a sound -- difference in volume, presence and volume of echoes or other time-delays, presence and volume of harmonics, and the particular characteristics of attenuation all factor in as well. Correspondingly, all these other clues change when sound reflects, not just phase.



Phase comparisons between tones are meaningful if and only if the tones are harmonically related; that is, their frequencies form a nice ratio like 2:1, 3:2, 5:4 etc. For those cases, it is just a mathematical exercise to measure phase relation. For all other cases, phase comparison is meaningless and mathematically undefined.



Insofar as it will preserve the acoustic events observed at the microphone to the limit of its resolution (frequency band and dynamic range), yes. There will, of course, be some effect on phase due to the physics of the listening device itself, but that's life.



Lossy compression means that it's fair game to do anything you want to a signal. Could phase information be lost? Definitely. It's quite likely, given the unspeakable things which a waveform endures in a lossy codec. However, in terms of its bearing on the quality of the finished product, it's not the biggest concern; other factors, like the wholesale elimination of entire frequency bands, will affect your perception of position much more strongly.

Please allow me to write a short discussion on what our auditory system captures.

At low frequencies, our auditory system uses phase difference (due to interaural time delay, otherwise known as ITD) to capture directional information.

At high frequencies, we use both ITD (but on the signal ENVELOPE, not on the signal itself) and ILD (that's interaural level difference) to distinguish direction. (head shadowing is very little at low frequencies)

We also use differences in timbre at the two ears (the HRTF "head related transfer function" or HRIR "head related impulse response" (really the same thing in another form) ) to determine front/back/up/down sorts of disambiguation along the cone of confusion (that's the surface that any particular ITD can arise from).

http://www.aes.org/sections/pnw/ppt/other/...d.ppt#256,1,The Science of Audio In 2005

Has some of this information in it.

I think the answer to this must be no. Given that we can tell whether a sound source is behind or in front of us, we're clearly using more than just phase to localize sound sources. Try plugging one ear - you can still tell which direction sounds are coming from, although it's somewhat more difficult.

So it's not just phase - and that means two speakers or two mics are never going to be enough. On the other hand a good stereo system can give a pretty good illusion of a sound stage, so it's at least not bad.

Indeed, although we only have two ears, they are directional devices, and as such, two channels of information are not sufficient to capture a soundfield, especially if one has a head that moves about, rather than stays stationary.

Head movement is really important - even if you think you're sitting still!

btw, lossy encoding can easily preserve transaural and binaural recordings, down to quite low bitrates. Unless you're using intensity stereo, the "phase" is fine. I've used mp3 at low bitrates, and though you can introduce horrible pre-echo and low pass filtering, the transaural and binaural effect (i.e. virtual source location) survives, but with artefacts on top.

Cheers,
David.

Agreed entirely!


Agreed again. Most of the strong binaural cues are well above threshold.

Has anyone seen that silhouette of a rotating dancer? It's been floating around the net.
For most people, once their brain decides to perceive she's rotating a certain way,
their brain then sticks with that direction. I think this example is a nice case study in
how the brain uses partial information: there really isn't enough, but the brain does what
it can, and we perceive more than we rigorously should.

In a similar way, the brain will form a strong opinion of where a sound is coming from, even if
the information used is imperfect. Hence, we get a good stereo impression from a set of
loudspeakers, even though the broadcast means cannot duplicate the original wavefronts.

Matt