Six Cardinal Rules of Sound Acquisition, Rev. 2

[Mallette]

For those of you interested in my recording adventures and explorations, a new revision of "Six Cardinal Rules of Sound Acquisition" is posted at www.mbsdar.com. With the new Beatles "White Album" project and some interesting findings regarding a possible methodology to produce reasonable surround from Redbook, I felt an update was in order.

Flames or comments welcome.

Dave

[pauln]

Well, I guess I'll start the questioning with some friendly fire...

It may take a moment to formulate my question because first I want to disclose my assumptions, some of which I am sure are incorrect:

From a geometric standpoint, I am assuming that with two ears one gets two signals with which to process and extract salient features that we hear is music and sound.

With one speaker mono, the channel reaches the ears at a slightly different time and this allows one to infer a first order lateral line of approach - but cannot distinguish whether the signal is from the front or the back. For example, the sound could appear to be coming from 30 degress right of center front or 150 degrees right from behind.

With two channel stereo, the above happens for each of the channels, but an additional second order level of this spatial inference occurs because each ear is getting both channels in which the sound source has been "placed" by the inate phase difference (pan) between the channels in the recording. So now each ear is getting four distinct versions of the sound for what was initially meant to be a single sound source originally recorded. The four sounds would be:

Left channel to left ear

Left channel to right ear

Right channel to left ear

Right channel to right ear

All this for a single sound like a lone flute.

And there are higher orders of the same from reflections and reverberation.

And none of this is yet sufficient to discern the sound source as coming from the front or the back.

Now getting to my question. I know that a sound coming from the front should sound different than from the back due to the shape of the head and ear, but I don't see how one makes the distinction to tell the direction of orientation front from back... If I hear a sound, I don't know in advance what it should really sound like from the front or the back. How is it that with just two ears it is really possible to isolate the true source of sounds coming from the front or back?

Bonus question: Does using four playback sources mean that each ear is getting 16 phase distinct versions of a single original sound?

[Mallette]

Sheesh, Paul, I am not looking forward to you getting ornery if this is "friendly" fire.

Actually, I am not going to get very deep into this as it is out of my league. I think I can state that my "gut" feeling is that, while the principles I've stated in the paper are hardly precise analogues of what happens in nature, they represent as close as technology is going to allow us to approximate for some time in the future. Obviously, sound comes at us from all directions, and in an enclosed space, from many different angles of incidence. There are some technical aspects of SoundCube that help to minimize phase confusion and supply the brain with a relatively simple triangulation problem that it seems to be able to work out pretty well. The brain seems to be able to fill in the blanks when required, hence our ability to enjoy the pancake flat moving images of a TV screen even though they are absurdly unreal...and will not fool my cat.

As to the ears, I believe them to have much in common with boundary or pressure zone microphones. The front advantage provided by the, danged if I can think of what to call them but you know what I mean, things is, I believe, designed to aid in location fixing and direction finding. Reversing yourself to the rear speakers in a SoundCube recorded sound field has a similar effect to doing the same in a real sound field, as does turning to the right or left.

The biological theory of this is, as I mentioned, WAY outside my abilities.

As to the bonus question, I really cannot say. I know that since the PZM's do not generate the common out of phase information that causes issues with other microphones that the potential for phase cancellation and the cranial confusion it generates is reduced. I'll let you count these and deduct them. PZM's produce no OOP info from anything that does not make its return within their 180 degree field of view. Does that reduce it to 8?

Dave

[pauln]

Aw hucks, I don't really know anything...

You're thinking of the "pinna", the latin for feather.

Actually, the 180 degree limit boundary of you mics would make quite difference with the orientation you use having an apex center forward.

[Mallette]

Aw hucks, I don't really know anything...

You're thinking of the "pinna", the latin for feather.

Actually, the 180 degree limit boundary of you mics would make quite difference with the orientation you use having an apex center forward.

You lost me, Paul. "...apex center forward?" You talking about the pinna thingy?

Dave

[Lonelobo]

1) Tell Dave it's never been recorded correctly but has the potential to have an audience in the tens and buy him a case of beer. Fin'

[oldtimer]

It has to be that the shape of the ear, relaying signals to the brain, is how we discern where sound comes from. That's why we turn our heads in the perceived direction of the sound to confirm. The recording made to duplicate hearing would have to take into account how much signal is perceived from which direction, and in what time alignment. The best way off the top of my head for this would be to take those antique RCA mics you have(well you only have one so you must get another), and configure an "ear" perceptive physical apparatus (shorten to ear-like thingy) or EPPA if you must, record, and then? Ahh, playback is hell right? Let's start by playing it back with pro logic IIx, so we will be at least sticking to some of the original theory. meanwhile, have another beer.

[Mallette]

Brian:
Good as it is, PLII is not a good substitute for the brain. And, while ribbons are my favorite mike, they are problematic as are other mike patterns as explained in the paper. I have eventual plans to deal with this, but right now I am sticking with the PZM's as I can get a perfectly overlapped 360 degree sound pattern with them without any direct reflection phase issues.

I use the 6 channel in (only 4 of them though, of course) input on my HT reciever for playback.

It is my belief that the ear functions as you suggest, and that is the foundation of Virtual Presence theory as presented. If you are in the center and turn your head, the image is also altered as in the real world. This doesn't happen with PLII as the steering is being done by the circuit rather than your brain.

Dave

[oldtimer]

Maybe playback on headphones is the answer? Assuming you record as the ear hears, then by reproducing through headphones you eliminate any uncontrollable room effects of the end user.

[Mallette]

All the end user has to do is stay inside the sound field.

In the quad days, Pioneer made a four driver headset that would work for VP. Don't know of any now...

Dave

[Mallette]

After reading the responses, I've decided Rev. 2 is still not properly explaining how the PZM and SoundCube work. I am working on Rev. 3 with more explanation and better illustrations. I'll post when it is up and ready for download.

Dave

[pauln]

Apex center forward was meant to refer to how you point the cube with a side edge most forward toward the front center.

No one is addressing my fundamental question - how do you tell if the sound comes from the front or the back with two ears?

Only if you already know how different a sound will sound whether it bounces or deflects in or off the ear at different orientations would the final signal coming from that ear have any chance of indicating the front back source direction (or above/below - really the whole ventral plane).

Of course that might be possible if the asymmetric outer ear masks the high frequencies a lot for sounds coming from the back and amplify those from the front, but then; does that mean a four mic and four speaker system would need to apply the low pass to the back channels and play the fronts louder?

Try putting you hands against the sides of you head like two giant ears, but place them in front of your ears so they point as if you had them on backwards - facing to the rear. Front sounds still sound like they come from the front! Even with the low pass operating in the wrong direction?

[Mallette]

>Front sounds still sound like they come from the front! Even with the low pass operating in the wrong direction?

Doesn't work that way, in my opinion, though I hypothesize you might fool a baby that way. The pinna (thanks for the word) provides the cue, but the brain does the work in a similar but far more sophisticated way as PLII. When I did what you suggested sitting here, my wife's air purifier behind me doubled in apparent volume...but remained in the rear. Why? 59 years of brain conditioning, IMHO. Toddlers I've observed don't immediately face a sound source, but search momentarily. If I cup my hands behind my pinna facing forward, directionality and volume increase to the front.

Actually, way back in the Vietnam war my stoned bungalow mate wanted to patent "acoustic headphones" that were basically two hands behind the ear as he thought it really enhanced the stereo image.

I remain convinced at this point that all directional cues must be provided by the brain to reproduce a space-time audio event accurately, and that in order to do so you've got to recreate the sound field that existed as precisely as possible. The PZM's with their resistance to out of phase audio and SoundCube's ability to provide a 360 degree pickup with near perfect mike overlap get as close as I know how to get to that.

Dave

[oldtimer]

Toddlers do that as part of the learning process, eventually they pick out more subtle clues. Unless they turn out to be drummers.

[Mallette]

Submitted for your approval at www.mbsdar.com, Rev. 2.1. This is actually a more substantial revision than 2.0, though no new ideas are involved. Please tear into it.

The mention of headphones brought up a couple of thoughts. A German company, Ultransone, builds headphones that deal with the binaural effect by directing the sound at the pinna rather than directly into the ear canal. However, this is still purely two channel even though they refer to it as "surround." It DOES point at a way to achieve transducer reversal as discussed in the SixCard. Pioneer made a set of 'phones during the quad era I believe would work if I could get a hold of a set. They used 4 transducers with 2 firing into the ear canal and the other two (LR, RR) firing from behind the pinna.

Dave

[Chris A]

In the mean time...

EDIT: This post was originally meant for another thread. Alas, it isn't possible to move this posting, but I think that the references to both reproduction and acquisition are synergistic properties...for your reading pleasure...

Abstract

Good audio sound quality in stereophonic reproduction may be summarized in the "eight cardinal points of loudspeakers," which are:

1) minimal distortion,

2) optimum size,

3) avoidance of rattles,

4) avoidance of shadows,

5) avoidance of cavities,

6) wide spacing,

7) proper number, and

8) toe-in."

[Don Richard]

No one is addressing my fundamental question - how do you tell if the sound comes from the front or the back with two ears?

I read something years ago that explained how sounds originating from behind the head, or above it, were located by the human psychoacoustic system. The ear does not have the same sensitivity to frequencies originating from behind, or above, the head as it does for sounds originating from the front. We learn over time that sounds with these frequencies out of balance are recognized by the brain as coming from the rear. An example that I remember is that a notch in the frequencies around 8 kHz will cause the sound to appear above the head and toward the rear.

That was part of it. The other part was arrival time differences between the two ears. The combination of these two conditions lead to accurate localization of the sound's source. I wish I could remember where I saw this because it contained experiments one could do that showed these phenomena. I tried some and they worked!

[Chris A]

I read something years ago that explained how sounds originating from behind the head, or above it, were located by the human psychoacoustic system. The ear does not have the same sensitivity to frequencies originating from behind, or above, the head as it does for sounds originating from the front. We learn over time that sounds with these frequencies out of balance are recognized by the brain as coming from the rear...

Toole's book is excellent on this subject. In fact he goes on to discuss the placement of surround speakers in a HT-type 5.1- or 7.1-arrangement. His point: speakers directly behind the listener do not increase the sense of "spaciousness" of the sound environment. He recommended something like 120 degree included angle as the spread (behind the listener) for surround sound effectiveness.

Chris

[100dB]

David, you've done some great theoretical and practical work here! Kudos. Localization is a combination of head shadow effects and pinnae effects (even torso effects) that result in something called Head Related Transfer Function (HRTF). You'll find lots of material on this topic from many sources. In addition, checkout Mapleshade Recordings - it will look quite familiar - PZMs on plexi (only two instead of four). Your pursuit of the "you-are-there" experience is what most of us classical engineers strive for. In your case though, you have control of both the recording and playback environments. Most of the time we have to make trade offs for headphone, two-channel, and surround formats. That can be tough. The "they-are-here" listening experience can be very cool too - using the technology to take the listener to an imaginary space. ...let's see, what else... Ribbon mics do indeed rule, but can be a pain; and I apologize for having no paragraph breaks - my iPod won't do it when posting. Oh yeah, if there is enough interest, I'll post a link discussing some research suggesting that higher sampling and bit rates yield absolutely no improvement in perceived resolution (but hey, memory is cheap these days so why not).

[Mallette]

Crikey, I think this is more response than the first time I put this up.

The reason I borrowed the "Cardinal Rules" thing from PWK was that it's in the same spirit. Granted, Paul felt that those who deviated from them were wrong. I don't agree with that, nor does the modern Klipsch organization. If that weren't true, the Paradigm would not exist. However, I do understand and agree with his basic point that what he describes is the simplest and most efficient means of reaching accuracy.

That's the point of the 6 Card. While there is absolutely nothing wrong with a fine movie track (or music) in surround produced by mixing and matrixing, then electronically parsing the material and steering it to a given speaker...it also isn't necessary or even desirable from a "purist" music lover POV.

SoundCube lives, but I've had almost no time to do further tests since I wrote that due to family and professional pressures, but I still intend to. One exception was a complete performance of the Beatles "White Album" done for charity a couple of years ago. It pointed to a very interesting technical application for Virtual Presence as an analytical tool. The event was held in a very large hall at the old Rice Hotel. Acoustics were pretty crappy. The amplification system was also pretty crappy, even though it was from a reputable event company. My goal was to record an acoustic space/time event...not the music. For one thing, I don't do non-acoustic recording, and for another I was more interested in recreating the event from a specific POV as a Virtual Presence experience.

In short, it worked swimmingly. The playback revealed a cacophony of boomy speakers, clanking glasses, fascinating discussions, and did so with completely precise surround positioning just as the paper suggests it should.

The event producer was pretty stunned. He said "It's weird to be precisely sitting in this room, but to close your eyes and be back at that performance. It also pretty disgusting, as we had no idea how totally muddy the sound was during the actual event. We have to fix that..."

So, it appears this technique is an excellent analytical tool for determining the qualities of an acoustic space.

Dave