3D audio report

[Gedlee]

I have been working with a prof at Princeton on a 3D system that he is developing. He asked for a set of Nathans on loan so that he could test them. He got back with me last week and told me that the Nathans were the best speakers that he had tested among dozens of other brand-name Hi-End loudspeakers. He was rather surprised that, according to him, only my speakers performed as the manufacturer claimed. He seemed surprised by that!

The link to the report is here:

3D audio report

PS - he has purchased the set that he has.

 

[mauidan]

I have been working with a prof at Princeton on a 3D system that he is developing. He asked for a set of Nathans on loan so that he could test them. He got back with me last week and told me that the Nathans were the best speakers that he had tested among dozens of other brand-name Hi-End loudspeakers. He was rather surprised that, according to him, only my speakers performed as the manufacturer claimed. He seemed surprised by that!

The link to the report is here:

3D audio report

PS - he has purchased the set that he has.

Watched the video. Based on his description and the narrow speaker placement, looks like he's doing something similar to Ambiophonics.

See:

http://www.ambiophonics.org

http://www.tactlab.com/Products/Ambiophonics/ambiophonics.html

What are the "dozens of other brand-name Hi-End loudspeakers" he compared your speakers to?

 

[Gedlee]

It is actually quite different than Ambiophonics, but thats not for me to discuss, its not my project. I only know of a few of the brands that were tested and it wouldn't be fair to mention those and exclude the ones that I don't know. It's better just to not name any of them - thats all I am comfortable with.

 

[kach22i]

Those speakers seem to be only 7" apart.

That must be one crazy program.

I'd like to experience it myself one day.

 

[mauidan]

It is actually quite different than Ambiophonics, but thats not for me to discuss, its not my project. I only know of a few of the brands that were tested and it wouldn't be fair to mention those and exclude the ones that I don't know. It's better just to not name any of them - thats all I am comfortable with.

I googled "Princeton University's Edgar Choueiri 3D Audio," it pulls up links connecting him with Ambiophonics.

"The development of ambiophonics is the work of several researchers and companies including Ralph Glasgal, founder of the Ambiophonic Institute; Dr. Angelo Farina, University of Parma; Robin Miller, Filmaker Technology; Waves Audio; Dr. Roger West, Soundlab; Dr. Radomir Bozovic, TacT Audio; and Prof. Edgar Choueiri, Princeton University."

http://www.ambiophonics.org/Bio.html

IMO, without a list of the "dozens of other brand-name Hi-End loudspeakers" he tested; Prof.Choueiri's statement doesn't mean very much.

 

[Gedlee]

Thats interesting, I didn't know that he was involved in Ambiosonics. At any rate what he is doing now is classic cross-talk cancelling. I guess he has some new algorithms that work well.

As to the other loudspeakers, lets just say that they were not cheap. None of the ones that he mentioned to me cost less than mine, so that puts some limits on them.

 

[Gedlee]

Those speakers seem to be only 7" apart.

That must be one crazy program.

I'd like to experience it myself one day.

In cross-talk cancelling, the closer the playback speakers are together the better the signal seperation becomes, but also the greater the errors due to limited processing resolution. I suspect that the Profs contribution is a higher resolution algorithm which allows the speakers to be closer together. With the higher directivity of my speakers he was able to improve the crosstalk seperation by 3 dB over any other speaker that he tried.

 

[Paul Spencer]

I stumbled upon the ambiophonics concept only a few days ago. Do you think we will see this enter home cinema the way EQ has? I find the concept of cross talk cancellation a bit hard to grasp. Any comments on how it works?

 

[RBFC]

The basic premise for crosstalk is that one ear hears information that is "meant" for the other ear, and hears it slightly delayed due to the longer path length to that ear from the speaker.

• So, the right speaker plays a sound. The right ear hears that sound, just as the right-side microphone in a stereo pair heard it.

• That same sound from the right speaker is also heard by the left ear, but slightly delayed because the left ear is a bit farther from the right speaker than the right ear.

• This "crosstalk" is what prevents more accurate imaging in both lateral location and depth perception, because the original sound is smeared by presenting the sound TWICE to each ear rather than once.

To cancel this crosstalk, a method similar to the "noise-cancelling headphones" used on airplanes is used. Engineers calculate the time delay for a sound from one speaker to hit the farther ear. They then create an exact opposite (inverted or 180 degrees out-of-phase) signal and play it from the speaker that is close to that ear. With the proper delay and an accurate inverted "cancelling" signal, the far ear cannot hear the opposite speaker anymore. So, the left speaker plays an inverted copy of the right speaker's signal, and plays it so that both speaker's signals arrive at the ear at the same time. Then, you cannot hear the signal, because it has been perfectly "zeroed out" at your ear.

That is the basic concept. Getting it just right is another story.

The IsoMike recordings from Kimber take a different approach to reducing crosstalk. They use a special baffle between the microphones during the recording session, so that each microphone picks up mostly only the sound meant for that channel, and very little of the sound meant for the opposite channel. So, they attack the crosstalk problem at the recording end.

I hope this helps. While I'm no engineer, this is basically how I understand the processes that have been employed in the past. I'm very interested to hear what Earl Geddes has to say about future implementations of the science to improve the effect.

Lee

 

[Kal Rubinson]

The basic premise for crosstalk is that one ear hears information that is "meant" for the other ear, and hears it slightly delayed due to the longer path length to that ear from the speaker.

• So, the right speaker plays a sound. The right ear hears that sound, just as the right-side microphone in a stereo pair heard it.

• That same sound from the right speaker is also heard by the left ear, but slightly delayed because the left ear is a bit farther from the right speaker than the right ear.

• This "crosstalk" is what prevents more accurate imaging in both lateral location and depth perception, because the original sound is smeared by presenting the sound TWICE to each ear rather than once.

To cancel this crosstalk, a method similar to the "noise-cancelling headphones" used on airplanes is used. Engineers calculate the time delay for a sound from one speaker to hit the farther ear. They then create an exact opposite (inverted or 180 degrees out-of-phase) signal and play it from the speaker that is close to that ear. With the proper delay and an accurate inverted "cancelling" signal, the far ear cannot hear the opposite speaker anymore. So, the left speaker plays an inverted copy of the right speaker's signal, and plays it so that both speaker's signals arrive at the ear at the same time. Then, you cannot hear the signal, because it has been perfectly "zeroed out" at your ear.

That is the basic concept. Getting it just right is another story.

The IsoMike recordings from Kimber take a different approach to reducing crosstalk. They use a special baffle between the microphones during the recording session, so that each microphone picks up mostly only the sound meant for that channel, and very little of the sound meant for the opposite channel. So, they attack the crosstalk problem at the recording end.

I hope this helps. While I'm no engineer, this is basically how I understand the processes that have been employed in the past. I'm very interested to hear what Earl Geddes has to say about future implementations of the science to improve the effect.

Lee

I do not know what the premises of Edgar Choueiri's work are because I have not been able to find any publications but I wonder about your analysis of crosstalk cancellation. In the real world, a violinist standing in front of me but off 35deg to the right is heard the same way by both ears and that contributes to my ability to localize it precisely. Put a mic there and what comes out of the right speaker works the same way. The fact that both ears hear both speakers is what distinguishes stereo listening over speakers from stereo (not binaural) listening over headphones.

Kal

 

[RBFC]

Kal,

Each ear gets an arrival when listening to a live source. When listening to stereo reproduction, each ear gets two arrivals: one from the closest speaker and a delayed one from the farther speaker. Remember, the microphones also had a delayed signal hit the farthest microphone... that delay is the auditory cue we need when listening to a live source. When the far speaker also is heard by the far ear, you get a second dose of the "delayed arrival" effect, so your right ear hears the original first arrival AND a double-delayed arrival due to the farther microphone/farther speaker complex.

What crosstalk cancellation attempts to do, in short, is turn the listening space into a giant set of binaural headphones. I remember playing with one of the early Carver Sonic Hologram Generators, which were a rudimentary example of the process I describe. When set up carefully, one could experience a full 180 degree soundfield that extended all the way to your sides, far beyond the speakers. Images had "reach out and touch it" realism. Tonal balance and richness of sound were unfortunately degraded somewhat, in the same manner that anaglyph 3D video flattens colors and brightness. However, the effect was startling at times.

Lee

 

[Bruce B]

Didn't Mathew Polk do this with his speakers? I believe they were connected together and there was some cancellation going on.

 

[terryj]

there are some free plug ins you can try.

this site has a lot of info on the topic

http://electro-music.com/forum/topic-15787.html

 

[Paul Spencer]

That makes sense, thanks. I saw the Ambiophonics website recently, but the mention of gaming made it sound like yet another gimmick. Now it's sounding a bit more serious and perhaps worth pursuing.

 

[Gedlee]

It works great when the recordings are done correctly. Problem is that it requires very specific recording techniques to work. There aren't many (any?) recordings like that, and I do not see the industry changing to doing so. It also begs the question - what is the role of a "recording" - to play back an "original space" or to creat a "new space" within the listeners space? Ambiophonics only really works for the former, it isn't really necessary for the later. I am not big on "recreating the impression of being in a music hall". Almost nothing that I listen has that as a goal. Virtually all studio creations use stereo as part of the art, not a window to the art. A lot of the disagreements that come up in audio stem from this basic point of view - what is it that you expect from your stereo?

It turns out that good speakers don't hinder either approach - thank God! My speakers were deamed to be the best availablke for cross-talk cancelling, because of their high directivity. But their design was mostly done simply to deliver what is on the recording to the listener - to create in the listeners space what the producer had in mind. Sometimes this falls flat, especially when "the room effect" is desired and it is not on the recording. In this later case people will often rate the lower "performance" speaker which has wide directivity and hence adds room ambiance (as spatiousness) as "preferred" because it adds something that the listener deems to be valuable. From a manufacturers standpoint, however, this is a dead end simply because the type of music that is enhanced by a "room effect" is simply becoming obsolete. Like it or not, "classical" music is declining, has done so consistantly year after year, and I personally do not see this trend changing. I like classical music to some extent, but I am more than satisfied to restrict my listening to it live. I get enough that way. I don't find it nearly as appealing in a home environment and as such I never listen to it at home. I am perfectly content to live with "classical music live" and "studio music" at home. This is very satifying to me. I sympathize with people who seek "live classical music" at home, because this will be a very small sector of the market. A sector however that is idealy suited to ambiophonics. So if "being there" is what you seek, then ambiophonics is for you, but it will be outrageously expensive and have a very limited reperatoire that is not compatible with standard stereo. The main stream will stay with stereo.

Oh yes, ambiophonics only works at a single seating location. I think the better choice for playback of this type will be insert earphones. I am doing a large scale study of these for a client and I'll tell you right now, no loudspeaker even comes close to them for "perfection" of the impulse response. This makes them ideal for ambiophonics.

 

[Kal Rubinson]

Each ear gets an arrival when listening to a live source. When listening to stereo reproduction, each ear gets two arrivals: one from the closest speaker and a delayed one from the farther speaker. Remember, the microphones also had a delayed signal hit the farthest microphone... that delay is the auditory cue we need when listening to a live source. When the far speaker also is heard by the far ear, you get a second dose of the "delayed arrival" effect, so your right ear hears the original first arrival AND a double-delayed arrival due to the farther microphone/farther speaker complex.

True and, on that basis, one might suggest canceling the right channel information coming through the left. OTOH, that information, if recorded properly, is necessary for stereo as opposed to dual mono since it captures some of the room-relayed sound present at the performance site. Deleting that might make completely lateralized sources a bit clearer but at the severe expense of any sources located between the mics.

What crosstalk cancellation attempts to do, in short, is turn the listening space into a giant set of binaural headphones. I remember playing with one of the early Carver Sonic Hologram Generators, which were a rudimentary example of the process I describe. When set up carefully, one could experience a full 180 degree soundfield that extended all the way to your sides, far beyond the speakers. Images had "reach out and touch it" realism. Tonal balance and richness of sound were unfortunately degraded somewhat, in the same manner that anaglyph 3D video flattens colors and brightness. However, the effect was startling at times.

Startling, I agree, but in my experience, unrealistic. I would describe it as hyper-realistic.

And, yes, Polk attempted similar with their M.A.R.S. arrangement. One hopes that we have moved beyond those relatively simplistic efforts, especially in this day of DSP. As an example of what can be done with it, I offer the Smyth Realiser that brings the experience of listening to a stereo or MCH room/system to headphones. That power can applied to signal encoding and probably is being applied, as we speak.

Kal

 

[RBFC]

Here's David Gilmour's newest CD, with a bonus CD encoded in 3D60, which supposedly provides a 360 degree soundfield from stereo systems:

http://www.amazon.com/Metallic-Sphe...r_1_18?s=music&ie=UTF8&qid=1286311136&sr=1-18

Lee

 

[Kal Rubinson]

I am not sure I agree with your hyper-realistic statement in full.

Here's why. One of the great features of my pre-amp is it is fully functional. (Hafler dh110). If you press the mono button on any "stereo" song you choose, and then after listening for a minute or so go back to full stereo, you might very well consider it hyper-realistic sounding.

Perhaps but that feeling would pass quickly as one re-adapts and would certainly be less of a disappointment than going the other way. Of course, the degree of difference would depend on the recording.

In fact, ture binaural is to my ears a much more superior method of playback than POS (plain old stereo!!). Anytime you use more than two pair (OK, add a center mike if you want) of mikes and then do ANY processing to the signal, you have created a new baby, and it aint the one at the live session. Its an illusion. I would like my illusion to better emulate the real thing, and image enhancement, cross talk reduction, whatever, if it makes it sound more like live, then I am for it.

Sure. A good binaural recording over phones neatly side-steps the downsides of POS but is an impractical alternative.

POS is really pretty poor all things considered. After 80 years we need a change.

Need I say that I could not agree more. I am on the record about this.

Furthering, IMO, as Earl was saying with classical music, POS really falls down in my opinion, it just can not cope with instruments in the dozens spread all over the place, while binaural can handle that because it transports you to the recording instead of trying to bring a faximilie of the recording to you.

As I said: A good binaural recording over phones neatly side-steps the downsides of POS but is an impractical alternative.

Kal

 

[RBFC]

Kal, how do you do those seperate quotes in the last post?

Tom

Tom,

At the bottom right of each person's post is a button with (")+ on it. This is the multi-quote function. Simply check the multi-quote button in each post that you wish to quote. Then, select "Reply with Quote" and use either the "Quick Reply" or the "Go Advanced" reply methods. You can insert your comments between each quote in the reply.

You can also manually multi-quote by inserting the V-bulletin code for quote before and after each section you wish to quote. You can PM me if you need more help.

Lee

 

[Kal Rubinson]

Kal, how do you do those seperate quotes in the last post?

Surround them with {QUOTE} and {/QUOTE} but use square brackets and not the curly brackets.

Kal