My old article on Loudspeaker Imaging

[damkor]

So, an analog system with better resolution...


...of high frequencies sounds better...no mystery. I thought he meant that the fractals added something, even though the final resolution of the digital information had a greater fidelity, by virtue of the rate of sampling, which was theoretically retreiving detail above the noise level of analog.

Best sound system I ever heard: SACD through 2W SETS and horns.

[dbe]

The transient response...


> ...of high frequencies sounds better...no
> mystery. I thought he meant that the
> fractals added something, even though the
> final resolution of the digital information
> had a greater fidelity, by virtue of the
> rate of sampling, which was theoretically
> retreiving detail above the noise level of
> analog.

> Best sound system I ever heard: SACD through
> 2W SETS and horns.

of a horn system with good low-wattage triodes is pretty startling, huh? Makes the hair on the back of my neck stand up just thinking about it ;-)

Dave

[romanbednarek]

Re: My old article on Loudspeaker Imaging


> Jeff Bagby
> Original 6/14/2001
> Revised 3/10/2007

Good stuff... I can't confirm everything you wrote but I can't find any fault in your logic. That's not a surprise though because I usually agree with most of the stuff you post (unlike the other Jeff B. that I know... my brother).

I can't comment much on the "microdetail" found in analog recordings because my experience with vinyl is limited to poor systems at a young age. However, I would like to comment on imaging as it pertains to speakers assuming that the recording is a constant in the equation.

First of all I would like to present my definition of imaging which is slightly different than yours. It is my understanding that imaging defines the ability to clearly pinpoint and localize sounds within the soundstage. Soundstage is the placement of those sounds within and beyond the speakers. I've found that there is often a bit of a tradeoff between soundstage and imaging at times where some speakers can have a very deep and spacious soundstage but a less defined image. I think this particular case relates to a bit of time smearing of the signal as I will discuss below.

I agree that reducing the amount of phase shift (group delay) produced by the crossover can be a key to providing better imaging and soundstage characteristics. In theory it makes sense because you are better preserving and reproducing the original signal without "smearing it in time." However, the tradeoffs with passive filters are often too great when trying to design a transient perfect speaker while preserving a flat frequency response and keeping the drivers within their optimal range. Digital filters should be able to overcome these sacrifices.

I would like to introduce a thought about drivers and how they might influence imaging and soundstage characteristics. You are probably familiar with my fascination with driver cone materials and how the physical properties might relate to sonic characteristics. One fellow proposed that a speaker cannot create a false sense of soundstage depth but I would have to disagree with this statement. I feel that drivers that have energy storage issues within the passband can possibly lead to a bit of time smearing of the signal and may lead to artificial soundstage depth. Also if the driver has CSD issues in certain frequency ranges the presentation of those frequency ranges may be skewed in the soundstage. I think this also can be related to how a driver transitions from pistonic to non-pistonic behavior as well. This relates to power response issues as well which is another factor that comes into play when you consider the room interactions and imaging. These are very loose theories but I think something might be going on when you consider different driver types and how they influence imaging and soundstage depth.

I think another key to imaging and soundstage is the tonal balance of a speaker. You mention that small speakers often have better imaging. Could this be due to the fact that the frequencies that are easily localizable by your ears are not being masked by the non-localizable bass frequencies of larger speakers (and doppler issues could also be a factor)? I'm convinced that putting a dip in the 1-3kHz region (or somewhere in that ballpark) can produce a less forward sound with less pinpoint imaging and maybe a deeper soundstage. With this in mind I feel that you must consider how the ear interprets different frequency regions in terms of localization.

Thank you for posting this topic. There are a lot of things that can't be proven regarding imaging but the more we can learn about it the better chance we have of designing better speakers or at least learning why a certain design sounds the way it does.

[dbe]

You said...


"I feel that drivers that have energy storage issues within the passband can possibly lead to a bit of time smearing of the signal and may lead to artificial soundstage depth."

***

Energy storage will indeed alter the soundstage, and will decrease the coherence of the microdynamics that produce spatial cues. Energy storage manifests itself as a distortion at the frequency that it is centered upon and the surrounding frequencies on either side of the storage frequency and (typically) at harmonic intervals of the center frequency.

Your categorization of soundstage and imaging are , ahem, spot-on ;-) They are intimately related but not the same thing. The soundstage presents the image, if you will. Soundstage and depth of image is partially defined by the distance from the side walls AND the rear wall. To get a stable, big soundstage and tight imaging also demands symmetry in the room, left to right. A room with greater distance on one side of the speaker will skew the image to the short sidewall side of the room.

Also, a system that has a very deep soundstage and less than optimal localization specificity probably needs to have the toe-in of the speaker adjusted. There is a critical balance of toe-in necessary for soundfield optimization. Sometimes this adjustment is as little as a degree or two of toe-in.

Another point that Jeff touched on is the Zero Delay Plane (ZDP) of the drivers in the system. This applies to 2-ways and 3-ways that cross the woofer below 400Hz. Correction of the time delay at the overlap of the crossover by physically tilting the speaker will retain the coherence of the crossover point by reducing the comb filtering and cancellation artifacts in the middle of the band of frequencies where the ear is the most sensitive. 1k to 3K is no place to have a crossover point unless the phase and time tracking is good at the crossover point. Phil Bambergs (PEB) speakers are some of the best imaging speakers that I have ever heard and I've heard hundreds. Why? Because Phil is anal about phase tracking at the crossover point.

Using Jeffs' PCD is a great way to see phase rotation before buying expensive crossover components. Phase tracking is one of the places that I am very careful to verify with measurements after the build. When the phase tracking is dead on the speaker sounds so.... right.

Speakers with dips or bumps in the 3K-6K range seem to have reduced specificity of both depth and localization to me. A very flat response in this range is critical for my ears.

My .02

Dave

[dbe]

Absolutely...


My pleasure.

DAve

[speakergeek]

Re: My old article on Loudspeaker Imaging


I appreciate the analysis but an understanding of imaging does not need to be muddied by concepts such as the fractal nature of sound. There are some very simple and concrete ways to understand the nature of audio image perception and in the process gain a wealth of understanding that can be applied to loudspeaker design and audio engineering in general. I will keep it short.

There is a way to perfectly record and playback sound with such a small loss of this information that the illusion of the original sound is complete. You can do this yourself with a few dollars worth of simple stereo equipment, and if done correctly, when played back, you will have such a strong sense of de-ja vu that it may be upsetting to you. What I am talking about is making your own binaurl recordings. This requires two small medium quality condenser microphones, small enough to fit into your ear canal, a digital recording device (laptop with sound card, whatever) and for playback a pair of good headphones. I have done this myself and the results are scary. You can hear sounds behind you, above, below, etc. I won't go into binaurl recording techniques here, you can find plenty of info on the web. The point is, it is possible to record and reproduce near perfect imaging with some inexpensive equipment and a pair of headphones, proving that all imaging information (phase information) can be preserved and reproduced in the recording chain.

This experiment proves that our perception of image is based on the phase differences of the audio wave at our ears.

So imaging is not a matter of analog vs. digital, vinyl vs. CD, Tube Amp vs. Solid State. Speaker design and room acoustics will have the most impact on imaging as can be proven by listening to you binaurl recording with headphones and then playing the same into a room with speakers. When the recording falls flat in the room, on stereo speakers you will wonder why.

If you do this and then spend some quality time thinking things through, armed with a little bit of info found on the web about how the brain locates sound in space, you will come to the conclusion that attempting to reproduce audio via stereo speakers in a room with any fidelity is a crap shoot. Most of the phase information will be destroyed by room reflections and the position of the speakers. For proof of this, lets do a thought experiment (or you can actually follow along at home). Lets say you have a pair of 4" full range drivers, in boxes, in the normal position for listening. With your binaurl recording playing back, the image will be approximately two dimensional. You will hear a left / right stereo field. Now take the speakers and (turning down the volume) position them as if they were headphones, close to your ears. Now the original phase information is correct, room reflections eliminated and you will be able to hear sound above, behind, in front, above and below. 360 degrees. From the same speakers.

There are ways to enhance stereo imaging (cleaning up room reflections, using dipoles, line arrays, etc.) but they are all compromises and illusions. The goal of a speaker designer is to second guess the environmental factors and human factors and come up with the best illusion. In my listening room, I prefer tall ribbon dipoles for this, but your mileage may vary.

[romanbednarek]

Re: You said...


> "I feel that drivers that have energy
> storage issues within the passband can
> possibly lead to a bit of time smearing of
> the signal and may lead to artificial
> soundstage depth."

> ***

> Energy storage will indeed alter the
> soundstage, and will decrease the coherence
> of the microdynamics that produce spatial
> cues. Energy storage manifests itself as a
> distortion at the frequency that it is
> centered upon and the surrounding
> frequencies on either side of the storage
> frequency and (typically) at harmonic
> intervals of the center frequency.

I think my interpretation is a bit different than this. I agree that a driver that "stops on a dime" after the signal is removed will more accurately reproduce the input signal and all of the microdynamics associated with it. However, I was under the impression that a driver that has tones that "linger" after the signal is removed will place those sounds deeper in the soundstage than other instruments which cover a different frequency range. That is what I mean by artificial soundstage depth. However, the problem that I ran into when I first introduced this theory was whether the time frame of that lingering tone was great enough to alter the spatial perception of that tone. Just to let you know where I got this crazy theory... I have worked with a few poly drivers in the past that seemed a bit too three dimensional in the midrange region where certain frequency ranges seemed to be placed deeper in the soundstage.

As far as the rest of your comments go, I am in total agreement.

[Jeff B.]

Re: My old article on Loudspeaker Imaging


> I appreciate the analysis but an
> understanding of imaging does not need to be
> muddied by concepts such as the fractal
> nature of sound. There are some very simple
> and concrete ways to understand the nature
> of audio image perception and in the process
> gain a wealth of understanding that can be
> applied to loudspeaker design and audio
> engineering in general. I will keep it
> short.

> There is a way to perfectly record and
> playback sound with such a small loss of
> this information that the illusion of the
> original sound is complete. You can do this
> yourself with a few dollars worth of simple
> stereo equipment, and if done correctly,
> when played back, you will have such a
> strong sense of de-ja vu that it may be
> upsetting to you. What I am talking about is
> making your own binaurl recordings. This
> requires two small medium quality condenser
> microphones, small enough to fit into your
> ear canal, a digital recording device
> (laptop with sound card, whatever) and for
> playback a pair of good headphones. I have
> done this myself and the results are scary.
> You can hear sounds behind you, above,
> below, etc. I won't go into binaurl
> recording techniques here, you can find
> plenty of info on the web. The point is, it
> is possible to record and reproduce near
> perfect imaging with some inexpensive
> equipment and a pair of headphones, proving
> that all imaging information (phase
> information) can be preserved and reproduced
> in the recording chain.

> This experiment proves that our perception
> of image is based on the phase differences
> of the audio wave at our ears.

> So imaging is not a matter of analog vs.
> digital, vinyl vs. CD, Tube Amp vs. Solid
> State. Speaker design and room acoustics
> will have the most impact on imaging as can
> be proven by listening to you binaurl
> recording with headphones and then playing
> the same into a room with speakers. When the
> recording falls flat in the room, on stereo
> speakers you will wonder why.

> If you do this and then spend some quality
> time thinking things through, armed with a
> little bit of info found on the web about
> how the brain locates sound in space, you
> will come to the conclusion that attempting
> to reproduce audio via stereo speakers in a
> room with any fidelity is a crap shoot. Most
> of the phase information will be destroyed
> by room reflections and the position of the
> speakers. For proof of this, lets do a
> thought experiment (or you can actually
> follow along at home). Lets say you have a
> pair of 4" full range drivers, in
> boxes, in the normal position for listening.
> With your binaurl recording playing back,
> the image will be approximately two
> dimensional. You will hear a left / right
> stereo field. Now take the speakers and
> (turning down the volume) position them as
> if they were headphones, close to your ears.
> Now the original phase information is
> correct, room reflections eliminated and you
> will be able to hear sound above, behind, in
> front, above and below. 360 degrees. From
> the same speakers.

> There are ways to enhance stereo imaging
> (cleaning up room reflections, using
> dipoles, line arrays, etc.) but they are all
> compromises and illusions. The goal of a
> speaker designer is to second guess the
> environmental factors and human factors and
> come up with the best illusion. In my
> listening room, I prefer tall ribbon dipoles
> for this, but your mileage may vary.

I appreciate what you are saying, and I understand well the physics behind how our ears localize sound this is precisely an area of study I focused on getting my Physiology degree. (Actually it quite a bit more complex than just path length differneces, but that is certainly part of it). I also understand how binaural recording preserves these phase relationships for us. But my article was really about how different design approaches in loudspeaker design could influence our ability to resolve this information better if it was already present within the recording. The fractal reference was mostly there to help paint the picture of the differences between the floor in analog and digital recording.

[robertcottiers]

Re: My old article on Loudspeaker Imaging


> I appreciate the analysis but an
> understanding of imaging does not need to be
> muddied by concepts such as the fractal
> nature of sound. There are some very simple
> and concrete ways to understand the nature
> of audio image perception and in the process
> gain a wealth of understanding that can be
> applied to loudspeaker design and audio
> engineering in general. I will keep it
> short.

> There is a way to perfectly record and
> playback sound with such a small loss of
> this information that the illusion of the
> original sound is complete. You can do this
> yourself with a few dollars worth of simple
> stereo equipment, and if done correctly,
> when played back, you will have such a
> strong sense of de-ja vu that it may be
> upsetting to you. What I am talking about is
> making your own binaurl recordings. This
> requires two small medium quality condenser
> microphones, small enough to fit into your
> ear canal, a digital recording device
> (laptop with sound card, whatever) and for
> playback a pair of good headphones. I have
> done this myself and the results are scary.
> You can hear sounds behind you, above,
> below, etc. I won't go into binaurl
> recording techniques here, you can find
> plenty of info on the web. The point is, it
> is possible to record and reproduce near
> perfect imaging with some inexpensive
> equipment and a pair of headphones, proving
> that all imaging information (phase
> information) can be preserved and reproduced
> in the recording chain.

> This experiment proves that our perception
> of image is based on the phase differences
> of the audio wave at our ears.

> So imaging is not a matter of analog vs.
> digital, vinyl vs. CD, Tube Amp vs. Solid
> State. Speaker design and room acoustics
> will have the most impact on imaging as can
> be proven by listening to you binaurl
> recording with headphones and then playing
> the same into a room with speakers. When the
> recording falls flat in the room, on stereo
> speakers you will wonder why.

> If you do this and then spend some quality
> time thinking things through, armed with a
> little bit of info found on the web about
> how the brain locates sound in space, you
> will come to the conclusion that attempting
> to reproduce audio via stereo speakers in a
> room with any fidelity is a crap shoot. Most
> of the phase information will be destroyed
> by room reflections and the position of the
> speakers. For proof of this, lets do a
> thought experiment (or you can actually
> follow along at home). Lets say you have a
> pair of 4" full range drivers, in
> boxes, in the normal position for listening.
> With your binaurl recording playing back,
> the image will be approximately two
> dimensional. You will hear a left / right
> stereo field. Now take the speakers and
> (turning down the volume) position them as
> if they were headphones, close to your ears.
> Now the original phase information is
> correct, room reflections eliminated and you
> will be able to hear sound above, behind, in
> front, above and below. 360 degrees. From
> the same speakers.

> There are ways to enhance stereo imaging
> (cleaning up room reflections, using
> dipoles, line arrays, etc.) but they are all
> compromises and illusions. The goal of a
> speaker designer is to second guess the
> environmental factors and human factors and
> come up with the best illusion. In my
> listening room, I prefer tall ribbon dipoles
> for this, but your mileage may vary.

SpeakerGeek: In the early seventies Sennheiser Corporation sold a modeled human head with high quality microphones embedded in the ear canals,It was designed to do Binaural recording,it was a bit out of my price range back then,I wonder if anyone makes a similar device today?..

I also had a binaural recordin(vinyl) which was put out by "Stereo Review magazine",it was an incredible listening experience with headphones.

Nice Post BTW
Bob C.

[speakergeek]

Re: My old article on Loudspeaker Imaging


> The fractal reference was mostly
> there to help paint the picture of the
> differences between the floor in analog and
> digital recording.

I was having a hard time understanding the fractal metaphor in the context of what you were trying to convey. I like to clarify and simplify, and I assumed this thread was about imaging, which we both agree has to do with preserving and reproducing phase information (path, difference, etc.) Using the binaurl experiment confirms that phase information is what matters when it comes to localizing sound. So you would get similar results regardless of the analog noise floor. You would just have well preserved phase (imaging) with a noise floor.

This understanding allows us to better analyze how to get better imaging from loudspeakers, by doing many of the things mentioned in this thread, such as preserving phase info by using digital crossovers, single drivers, better passive crossover design, eliminating box resonances, treating room reflections (a big factor), etc.

But lets be clear about what is responsible for our perception of imaging - brain interpretations of phase differences between our 2 stereo ears. Understanding that, we can work with tools and components to come up with speakers that image better.

I think you have done a great job of sparking interest and conversation about this subject and I think that is what this board should be about, and why I jumped in. Kudos for sticking yourself out there and being a gentleman. Hopefully we will all keep learning.

There are many other factors that help us perceive sound to be real, most of them touched on here, but I am sticking with imaging which may not be as well understood as some of the other factors, such as noise, transient response, frequency response, harmonic distortion, etc. All good topics of study.

[speakergeek]

Re: My old article on Loudspeaker Imaging


> SpeakerGeek: In the early seventies
> Sennheiser Corporation sold a modeled human
> head with high quality microphones embedded
> in the ear canals,It was designed to do
> Binaural recording,it was a bit out of my
> price range back then,I wonder if anyone
> makes a similar device today?..

Yes, there are quite a few of these "head" type microphones, but there are also headset types that are much cheaper and allow you to record freely using your own head (so to speak). The results are just as good, and if you play back your own recordings to yourself, de-ja vu.

> I also had a binaural recordin(vinyl) which
> was put out by "Stereo Review
> magazine",it was an incredible
> listening experience with headphones.

You can find some samples of binaurl MP3s on the web, and even with MP3 compression (uck, for us audiophiles) the phase information is retained and the illusion is complete. This illustrates my point that phase information is independent of other factors such as noise floor, transient response, etc. when it comes to localizing sound.

[unknownuser]

Vinyl Theories and other interesting diversions


Hey Jeff,

How goes? What a huge topic. I could go on for hours, but I only have minutes available so….

Hate to be the party pooper but (you knew this was coming, given I’m a PITA some times J ):

>My theory is based on the fact that psychoacoustically we perceive the size of a sonic image by the relative intensity of the ratio of direct and reflected sound and the delay between the two.

While not diminishing the importance of the spectral balance of early reflection, or of getting the right balance between early and late, unfortunately this theory isn't really true. The relative size is dependant more upon the IACC, Inter-Aural Cross Correlation. Check out Greisenger's web site, or google envelopment (eg <A HREF="http://www.concerthalls.unomaha.edu/...n/spacious.htm">http://www.concerthalls.unomaha.edu/...n/spacious.htm</A>).

For example, you could generate massive images in an anechoic chamber, using Sonic Hallography type approaches from 2 channels. FWIW, a neat experiment: wire one tweeter (say, left) out of phase with the other (say, right). The image gets massive. Height reaching to the skys etc.

A few of my own thoughts on why analog images better than CD:
- lower SNR: the noise helps us perceive the low level info (before I get harassed about this by anyone, look it up)
- lower cross talk. This I think is "The Kicker". By definition, stereo itself is horribly flawed by the crosstalk from right speaker to left ear and left speaker to right ear. It is my theory that LP's notoriously poorer crosstalk than digital actually aids in retrieval spatial info by: creating more, by its own inherent crosstalk; through the actual lack of IACC in the crosstalk, which adds spaciousness.

I’ve never seen this theory floated before, but its been banging around in my head for years and I think it has real technical merit to it.

The fractal analogy is interesting when describing masking, but its an analogy only. Making is masking, there’s no inherent level resolution limit to digital that means it struggles more with masking. You could argue that digital has higher resolution, based on its much superior SNR (especially with chaotic dither: see Shannon theory).

Why the Rega was flatter than the Linn, I dunno, but I know exactly what you mean. FWIW, the most outrageous (good) depth I ever heard from point sources was the ancient Rega 2+Grace arm +Grado cartridge over the JR149s (now you know why I’ve slaved over those B110 bitches for so long).

Boy do I ever agree about stiff cabinets being better. I've irrevocably proven that to myself through experimentation and measurement. A dead cabinet sounds far more real, less fatiguing, there’s no comparison. Good lessons there.

I also think the “smaller images better” gets back to that old power response though; A big speaker will just struggle in that area more. “Flat power response” is a misleading target. You want not just the sum of all reflected powers to hit a certain smooth target, but individual reflections need to look “right”, spectrally. This is where small is beautiful and large is the 1000lb gorilla in the room. Small is closer to point source and can give smoother individual reflections.

I found the exact same thing as you did Jeff, that single drivers (or itsy bitsy cabinets) image like gang-busters. My theory: see the preceding paragraph.

BTW, Sonic Hollography cooked up some amazing effects didn’t it? My TV had it and it could place images in the neighbours yard next door. I kid you not: the virtual image would walk from the yard, through my wall, through the speakers, back out the other side and into the other neighbours wall. Just like it was supposed to on the TV. What a rush!

BTW2: I tried but never had the same success with low group delay systems. Never noted a great imaging improvement. Could be my set up limiting it. I prefer low order xovers because they tend to stress the tweeter less (as long as it goes to high order out of band).

Wish I could go on (and on), imaging was a pet professional topic of mine for a long time, but duty calls.

PS To Ron E, you know this, depth is created by a back wall reflection, mainly. Simple mirror ray math. Not to knock your points though, I really appreciate you as a voice of reason. The ability to hit the back wall with the right signal (dipoles anyone?), without clouding the picture with diffraction etc, is key.



(Originally posted by: DDF)

[Jeff B.]

Re: My old article on Loudspeaker Imaging


"This illustrates my point that phase information is independent of other factors such as noise floor, transient response, etc. when it comes to localizing sound."

I think you are missing my point, or I am not communicating well. I agree with you 100% that localization is mostly based on delay (not really phase differences) between the two mics. I am just saying that in order to extract that information well we need to minimize the affects of these other things, for this is low level information in many recordings.

[speakergeek]

Re: My old article on Loudspeaker Imaging


> "This illustrates my point that phase
> information is independent of other factors
> such as noise floor, transient response,
> etc. when it comes to localizing
> sound."

> I think you are missing my point, or I am
> not communicating well. I agree with you
> 100% that localization is mostly based on
> delay (not really phase differences) between
> the two mics. I am just saying that in order
> to extract that information well we need to
> minimize the affects of these other things,
> for this is low level information in many
> recordings.

I agree that many recordings have not preserved the phase information, so things like dipoles will bring the illusion back to life (back wave out of phase with front wave).

If the phase information is in the recording then the weak link is the speaker and the room, not the electronics (assuming the electronics and crossover don't mangle the phase either). Minimizing room reflections directly behind the speakers is a start. Designing speakers so off axis radiation is smooth helps with imaging. Line arrays eliminate vertical radiation that will bounce of ceilings, etc.

I think delay between two signals that are otherwise identical IS phase information, so we are talking about the same thing. In the recording studio to simulate stereo from a mono signal engineers will sometimes just split the mono signal, pan one left and the other right, and delay the right channel by milliseconds so it is out of phase by 90-180 degrees. You now have the right channel out of phase with the left and a stereo image. The mono signal will float somewhere between the two speakers.

Your ear/brain interprets the delay / phase shift as a directional sound with an 'image'. It just gets complicated when you start talking about many paths of the reflected sound, many delays, a multitude of phase shifting. That is what happens with room reflections. Your brain knows the difference. It knows you are in a room, because the reflections and phase information tell it so. It is very hard to fool your brain into thinking otherwise.

Using a binaurl recording with headphones eliminates all room reflections and preserves all of the original phase information. How can we get stereo speakers to do that, especially when the original recording was a multi-track, or mono mic'd like most modern recordings? So that is the trick, the illusion, that is pulled off to varying degrees by combinations of recording and playback electronics, speakers and rooms.

[Jeff B.]

Re: My old article on Loudspeaker Imaging


> SpeakerGeek: In the early seventies
> Sennheiser Corporation sold a modeled human
> head with high quality microphones embedded
> in the ear canals,It was designed to do
> Binaural recording,it was a bit out of my
> price range back then,I wonder if anyone
> makes a similar device today?..

> I also had a binaural recordin(vinyl) which
> was put out by "Stereo Review
> magazine",it was an incredible
> listening experience with headphones.

> Nice Post BTW
> Bob C.

The ear-brain localization funtion is based on two primary mechanisms - the difference in path length delay that sounds take to get to the two ears, and what is called the Head Related Transfer Function of the sound. This is a fancy way to refer to the fact that our ears change the frequency response of sound depending on the angle sound enters it from, and low pass function of our head. This is caused by the fact that the shorter wavelengths of high frequency sound cannot wrap around an object the size our heads. Consequently, the amount of high frequency information that's present will be very different between our left ear and our right ear for sound coming from our right side, for example. Our brain uses this change in high frequency information and the differences in the delay in the sound between both ears to determine the location that the sound is coming from. The dummy with the mics in its ears that you refer to was designed to model this as closely as possible. Interesting , huh? By the way, we have one of these at work too.

Jeff B.