BACCH-SP: The future of high-end audio? Yes.

[soundArgument]

In January, I asked on WBF whether the BACCH-SP--an audiophile product that aims to implement the findings of Prof. Edgar Choueiri--represents the future of high-end audio:

http://www.whatsbestforum.com/showthread.php?16640-BACCH-SP-the-future-of-high-end-audio

Prof. Choueiri is, by all accounts, a music lover, audiophile, and analog enthusiast. His research team at Princeton University has studied how to retrieve fully the three-dimensional spatial information locked within stereo recordings. The way to retrieve this information, the professor claims, is to eliminate the crosstalk that occurs during playback of stereo recordings. Some crosstalk between channels results from the use of stereo miking techniques; this type of crosstalk is desirable for stereo imaging. But unlike crosstalk that results from using stereo miking, the crosstalk that occurs during playback over stereo speakers is additive. That is, when the sound from your right speaker reaches your left ear and vice versa, you're hearing something that wasn't on the recording. The BACCH-SP uses processing of extraordinary sophistication to eliminate the additive crosstalk that occurs during playback. As we all know, however, sophistication doesn't entail effectiveness, so I've been curious about whether the system works.

On a recent visit to Hong Kong, I had the pleasure of meeting, and hearing the systems of, two users of the BACCH-SP. I can confirm that the device does represent the future of high-end audio. I've heard a lot of systems, including supersystems (at shows and elsewhere) and highly revealing studio monitoring systems. The two BACCH-SP systems reveal more of what's on recordings than any other systems that I've heard. Listening to them was also addictive. I did not want to leave either listening room, nor did I want to go back to listening to music without the BACCH-SP.

___

The first system I heard is housed in a carefully treated listening room in the office of Mr. WS Lam's firm MASIS Audio. The system consisted of Sanders electrostatic speakers, Sanders amplifiers, a Velodyne powered sub (no idea what the model was), and pro audio cabling (mostly Mogami, I believe). (The Avantgarde Zeros pictured were on static display.) The source was a Mac-based music server playing files at various resolutions. Mr. Lam's assistant calibrated the BACCH-SP by placing two intraaural microphones in my ears and running test tones as I leaned to the left and to the right. A head tracker connected to the BACCH-SP continuously monitors the position of the listener's head, ensuring that the device's three-dimensional effect can be heard within a sweet spot that is several feet wide. The assistant then demonstrated that the BACCH-SP was working by leaving the microphones in my ears for a moment, playing a song on his smartphone's speaker, walking in a circle around me, and recording the sound through the intraaural microphones. The assistant then played the recording back on the BACCH-SP, and indeed, I could hear the song moving around my head.

Mr. Lam then came in, and started playing music. Using the BACCH-SP, we listened to a variety of commercially available recordings, mostly of classical music. As Mr. Lam explained, binaural recordings result in the most realistic three-dimensional sound reproduction through the BACCH-SP. A number of recent Chesky recordings use the binaural recording technique, and indeed, these recordings sounded startlingly realistic, with sounds seeming to emanate from far beyond the positions of the Sanders speakers.

To my surprise and delight, non-binaural recordings of classical music also sounded remarkable, if slightly less three-dimensional, through the system. We listened to a number of recordings on the EMI, Decca, DG, and Reference Recordings labels. On orchestral recordings, I could clearly hear woodwind sound coming from behind strings, and the string sections seemed to extend laterally far beyond the positions of the Sanders speakers. The relative heights of percussion and brass instruments were clearly audible as well. Chamber music recordings were, if anything, even more impressive, with the scale of chamber ensembles clearly distinguishable from that of orchestras.

The effect of the BACCH-SP in this system was not subtle. Mr. Lam toggled a bypass switch several times, triggering a return to conventional stereo imaging, which sounded both one-dimensional (left-right only) and homogenous by comparison.

This demonstration was extraordinarily impressive and might represent the best stereophonic sound currently possible. I was curious to hear what the BACCH-SP sounded like in a less ideal environment.



___

The second BACCH-SP system was in an audiophile's relatively large Hong Kong apartment. The speakers were models of the audiophile's own design, apparently inspired by the Rockport Arrakis, but using digital crossovers and room correction. Amplifiers were Ncore models, and, again, the source was a Mac-based music server. All the digital gear was slaved to an Antelope master clock. In this system, the BACCH-SP was not calibrated to my ears. We simply started listening to music. I sat in the center sweet spot, and the audiophile owner sat a few feet to my right.

Despite these less ideal conditions, the effect wrought by the BACCH-SP was still obvious. The soundfield extended far beyond the positions of the speakers. The depth of the stereo image exceeded what I've heard through any conventional system. The effect of the BACCH-SP was obvious despite the presence of more than one person in front of the device's head tracker and in the absence of specialized room treatments.

We listened to a few binaural recordings (I believe these were all Chesky recordings but am not certain), and the effect was stunning, with instruments seemingly arrayed in space around the listening position. Conventional recordings of classical music sounded spectacular, too. I've never heard the interpretations of Kleiber, Abbado, and Solti sound so three-dimensional. I thought I knew these recordings, but only the BACCH-SP allowed the spatial information on them to be revealed fully.

I should add that even without the BACCH-SP, this system might have been one of the best I've ever heard. The speakers have extraordinary dynamic capabilities and low-frequency extension. With the BACCH-SP, the system produced a three-dimensional soundfield of greater realism than that generated by any other system I've ever heard, with the exception of the one in Mr. Lam's office.



__


Strangely, the BACCH-SP can be so revealing that it might not solve all problems: it seems that some studio recordings engineered for playback on conventional systems or Apple earbuds probably sound better without it. For recordings made in natural acoustic spaces with stereo miking techniques, however, the BACCH-SP allows the listener to hear information that was previously inaccessible. It's for real.

The technology seems to have attracted a lot of admirers, including David Chesky, and the audiophile press seems to have been wowed by the device at CES. No one's exaggerating how effective the BACCH-SP is. It's revolutionary, and it makes the improvements in playback wrought by a number of other technologies (e.g., DSD playback) seem rather trivial by comparison.

Have I bought one? No, I haven't, but only because I can't responsibly spend $55,000 on audio right now. However, unlike the value of many audiophile products, that of the BACCH-SP seems to me to be indisputable. I hope more serious music listeners can hear it, and I plan to buy one when I can.

 

[esldude]

So Bob Carver's Sonic Holography lives again.

That is basically what his system did. It too, with some recordings, in the central position, with careful tweaking could give effects like you described. Once heard it with some of Carver's pseudo-panel speakers, and it seemed with minimally miked recordings to fill a 30 foot wide space with speakers only 8 feet apart and 8 feet distant.

If you read the old patents from Blumlein, he was aware of the cross-talk with speakers. But despite the crosstalk the speakers create a time delay between channels. It is also why in some sense spaced pair recordings aren't really right for speaker listening (good over headphones). You get too much interchannel delay.

Oh well, an update of that which this system appears to be might well be worked out better than in the past. I would imagine however, based upon the experience of the old Carver version, multi-miked, processed, synthetic recordings are going to be a mixed bag.

I do realize this system does more than Bob Carver's old system. Calibrating for your ears, and adjusting the effect for movements of your head would be a big extra. Still in principle it is the same idea.

 

[Fitzcaraldo215]

I agree this appears to be Bob Carver's Sonic Holography (BCSH) on steroids. The big, big question is what the size of the sweet spot and what happens outside it in the listening room. BCSH was not good in that department, and I suspect a similar issue here. Personally, I believe I will continue to favor hi rez multichannel listening with discretely recorded Mch material, which I am deeply into and greatly love.

I am somewhat limited by repertoire for discrete Mch, but not so much since I favor classical music. This device has the advantage of retrieval from stereo sources, but there are plenty of other Mch from stereo synthesizers if I wish to listen that way, which I generally do not. Maybe this is better than alternative synthesizers, but that is not something I am dying to have.

It is also worth noting the parallels between this system an the headphone oriented Smyth Realizer, although the Smyth is not really a spatial enhancement from stereo synthesizer. But, the IR head position sensing and the ear bud mike calibration ideas have similarities.

 

[Kal Rubinson]

I thought it a less comfortable experience than with Smyth. I also had the feeling that the spatial expansion was strangely consistent with a range of recordings from widely disparate sources.

 

[dallasjustice]

I asked and was told there's no plan for a software-only solution. (eg. Plugin or convolution engine). Expensive box DSP solutions always seem very myopic and gimmicky to me.

 

[soundArgument]

I thought it a less comfortable experience than with Smyth. I also had the feeling that the spatial expansion was strangely consistent with a range of recordings from widely disparate sources.

I didn't hear the homogeneity to which you refer, and I spent six hours listening to the BACCH-SP on two separate systems. Isn't the Smyth Realiser intended to achieve a totally different effect--i.e., replicating the experience of listening to conventional stereo speakers on headphones?

I do understand the sense of discomfort to which you point. In the treated room with Sanders speakers, the BACCH-SP made the system sound utterly different than anything I've ever heard. On the system in the conventional living room, the effect was less pronounced and the experience more similar to listening to a conventional (non-BACCH-SP) system. I guess I'd say that, to my ears, the domestic system sounded more comfortable because, to a greater degree than the other system, it resembled what I'm used to hearing.

But the system in the purpose-built room, I'm almost certain, is the most accurate system I've ever heard.

 

[soundArgument]

I asked and was told there's no plan for a software-only solution. (eg. Plugin or convolution engine). Expensive box DSP solutions always seem very myopic and gimmicky to me.

I can see at least two reasons for the hardware DSP approach in this case. The first is that the investment in this technology was huge, by audio gear standards, and a software plugin would be vulnerable to piracy.

The second is that the effect of the processing cannot be consistently achieved without head tracking. According to Mr. Lam, use of the head tracker requires the BACCH-SP to process a very large number of streams of audio in parallel, using a lot of CPU horsepower. Consequently, even if one could get a software version of this technology and a head tracker to connect to the computer, the computer might not be powerful enough to run the processing and JRiver, Audirvana, etc. simultaneously.

 

[dallasjustice]

I agree about the head tracker. But that's the only piece of hardware which may be needed. Any modern computer will be WAY more powerful than whatever FPGA is inside the BACCH-box. Have you ever checked to see how much CPU is used while playing a file in Jriver, for example? It's usually around 1% on my server.

Piracy is very unlikely especially since the head tracker is required and simple safeguards can be taken. Has anyone ever pirated any other DSP software?

Margins on software are much higher than hardware. It probably wouldn't be a plugin. It would need to be a stand alone convolution engine and measurement setup software.

The other problem is bass. To me, bass matters a lot and the BACCH seems to limit bass possibilities.

It is easier to market, sell and setup a box like this. That's the reason for the box.

I can see at least two reasons for the hardware DSP approach in this case. The first is that the investment in this technology was huge, by audio gear standards, and a software plugin would be vulnerable to piracy.

The second is that the effect of the processing cannot be consistently achieved without head tracking. According to Mr. Lam, use of the head tracker requires the BACCH-SP to process a very large number of streams of audio in parallel, using a lot of CPU horsepower. Consequently, even if one could get a software version of this technology and a head tracker to connect to the computer, the computer might not be powerful enough to run the processing and JRiver, Audirvana, etc. simultaneously.

 

[soundArgument]

I believe the processor in the BACCH-SP is a high-end Intel CPU, not an FPGA.

I agree about the head tracker. But that's the only piece of hardware which may be needed. Any modern computer will be WAY more powerful than whatever FPGA is inside the BACCH-box. Have you ever checked to see how much CPU is used while playing a file in Jriver, for example? It's usually around 1% on my server.

Piracy is very unlikely especially since the head tracker is required and simple safeguards can be taken. Has anyone ever pirated any other DSP software?

Margins on software are much higher than hardware. It probably wouldn't be a plugin. It would need to be a stand alone convolution engine and measurement setup software.

The other problem is bass. To me, bass matters a lot and the BACCH seems to limit bass possibilities.

It is easier to market, sell and setup a box like this. That's the reason for the box.

 

[dallasjustice]

I believe the processor in the BACCH-SP is a high-end Intel CPU, not an FPGA.

Ok?

So it wouldn't be a big leap to offer the software instead of the box which would greatly reduce the cost and increase the profit.

 

[soundArgument]

Ok?

So it wouldn't be a big leap to offer the software instead of the box which would greatly reduce the cost and increase the profit.

I was just trying to make the point that the processing may actually use a lot of CPU, more than one would imagine.

In any event, I just hope others can hear recordings played back using the type of processing currently available only through the BACCH-SP. It's the first transformational audio innovation to arrive during my lifetime.

 

[dallasjustice]

I am 100% in favor of new technologies in audio and I am a DSP enthusiast for sure. However, it's always disappointing to me to see a potential new advance be muted by factors having nothing to do with technology.

Btw, the concept behind BACCH isn't really new. Crosstalk cancellation DSP has been around a while. Whatever BACCH is doing different from before, I hope a much larger segment gets the chance to demo it.

I was just trying to make the point that the processing may actually use a lot of CPU, more than one would imagine.

In any event, I just hope others can hear recordings played back using the type of processing currently available only through the BACCH-SP. It's the first transformational audio innovation to arrive during my lifetime.

 

[Kal Rubinson]

I didn't hear the homogeneity to which you refer, and I spent six hours listening to the BACCH-SP on two separate systems. Isn't the Smyth Realiser intended to achieve a totally different effect--i.e., replicating the experience of listening to conventional stereo speakers on headphones?

Certainly. I only compared them because both are attempts to present a surround field with two channels of transduction.

I do understand the sense of discomfort to which you point.

Part of my discomfort was my implicit feeling of constraint in moving about.

But the system in the purpose-built room, I'm almost certain, is the most accurate system I've ever heard.

I did not find it so in comparison to the best of discrete multichannel reproduction. I found the soundstage to be more spacious, especially with its ability to expand vertically and to represent proximity, but it suffered in its inability to disassociate individual voices/instruments from the general ambiance.

 

[soundArgument]

I did not find it so in comparison to the best of discrete multichannel reproduction. I found the soundstage to be more spacious, especially with its ability to expand vertically and to represent proximity, but it suffered in its inability to disassociate individual voices/instruments from the general ambiance.

I've never heard the best multichannel reproduction, but on the BACCH-SP systems, I thought that the differentiation of instruments from the surrounding acoustic spaces was remarkable. For instance, on conventional systems, chamber music tends to sound spatially problematic: a small group of performers often sounds unnaturally wide, and this problem is often accompanied by poor image stability.

In contrast, on the BACCH-SP systems, chamber music recordings (even some badly done recordings that I know well) had appropriate spatial perspective, each with a small group of performers in the center of the soundfield and hall reflections surrounding the listener. String quartets and violin sonatas were presented though those systems as I've never heard them.

 

[Atmasphere]

I've never heard the best multichannel reproduction, but on the BACCH-SP systems, I thought that the differentiation of instruments from the surrounding acoustic spaces was remarkable. For instance, on conventional systems, chamber music tends to sound spatially problematic: a small group of performers often sounds unnaturally wide, and this problem is often accompanied by poor image stability.

Every time I read comments like this, I wonder if Blumlien is known to the person offering the technology. But also, I wonder if the person writing the post might have some sort of setup problems: 'poor image stability'? I've yet to hear something like that on my system!

The whole point of the Blumlien system is 3 dimensions, which is does quite well, exhibiting all the properties also attributed to this system. Pardon me for asking and no offense intended, but do you understand how stereo is supposed to work?

 

[Brucemck2]

The guy who developed the technology is both a hard core audiophile and a real rocket scientist.

http://www.princeton.edu/3D3A/Choueiri.html

I spent some time with him late last year on the phone, and he clearly has a "theory of the case" based on sound (no pun intended) mathematics/physics concepts. What really impressed me though was his willingness to help me properly integrate room/speaker correction filters into his processing scheme, even as a one-off case.

 

[soundArgument]

Every time I read comments like this, I wonder if Blumlien is known to the person offering the technology. But also, I wonder if the person writing the post might have some sort of setup problems: 'poor image stability'? I've yet to hear something like that on my system!

No, no setup problems. Being a child of two music professors and a former aspiring classical musician, I've attended hundreds of performances of classical music, and continue to attend concerts. String quartets and violin sonatas tend to be reproduced poorly by all conventional stereo systems (even ones with high-powered output transformerless tube amps): the spatial relationships simply are not right, and everything tends to sound too big, too spread out. The BACCH-SP mitigates this problem, even on especially problematic recordings. An example is the DG Kremer/Argerich recordings of the Beethoven sonatas, some of which I heard on Mr. Lam's system and which I thought the BACCH-SP improved drastically. I digress, but it was nice to hear great performances finally get the playback quality they merit.

The whole point of the Blumlien system is 3 dimensions, which is does quite well, exhibiting all the properties also attributed to this system. Pardon me for asking and no offense intended, but do you understand how stereo is supposed to work?

Indeed, I do! I spent a little time as a (paid) engineer of recordings of classical music before moving on to other things, and I'm familiar enough with stereo miking techniques to know that 'Blumlien' is spelled Blumlein (sorry, just sayin') and also doesn't make sound blossom into three dimensions when played back through a conventional stereo.

The purpose of the BACCH-SP is to allow the spatial relationships on the recording to be fully heard. Putting speakers in a room causes you to hear many distortions--principally the crosstalk between the speakers but also reflections from the walls, etc.--that the theoretically perfect stereo system would not cause you to hear. By mostly eliminating the crosstalk between the speakers and mitigating some other distortions with very precise EQ, the BACCH-SP represents an advance toward the theoretical ideal.

I realize it isn't for everyone. (I'm surprised the august Dr. Rubinson didn't like it, but I don't hold it against him.) It's a big change for most audiophiles to accept digital processors that alter the sound, and as the BACCH-SP's supporters admit, it doesn't work on every recording.

Still, it's amazing, and you have to hear it.

 

[adyc]

After seeing this thread, I made a booking with MASIS audio for the demonstration of BACCH-SP. I spent around one hour and an half there. There is no exaggeration. It indeeds changes one mind about conventional stereo listening.

The first demonstration is binaural recordings of a guy in a cathedral walkings toward you 30 feet away from your right side and whisper to your right ear. The most unreal part is that you hear that guy is really whispering to your right ear. Without the filter, the whispering happening in front of your face. It is not imagination. I switched on and off the filter and the guy whispering switching from right ear to front of face with filter on and off. The demonstration continued with more binaural recordings of some virtual barber shop which demonstrated a barber cutting your hair around your head.

Enough with the binaural recordings, I asked how BACCH-SP works with conventional stereo recordings. The first track is Pink Floyd's Money. With the filter on, the sound of cash register is really happening right next to right and left ears. However, I do not like the effect too much. I understand it may be due to the mixing. However, to me it is not real. You do not go to concert with the music playing to your left and right years. So I asked them to play some chamber music. The track is Bartok String Quartet No.1. With the filter on, you can clearly differentiate first and second violins. Furthermore, the first violin is extended beyond the boundary of the speaker. I asked to bypass the filter. The sound stage collapsed within the speakers and first and second violins mixing with each other. The demonstration continues with more conventional stereo recordings. And indeed BACCH-filter works really well.

However, I found the sound of Sander speakers is very thin and do not have much body. This is not a criticism of BACCH-SP because I hear the same thinness with the filter off. Unfortunately, BACCH-SP requires high directivity speakers to work. My speaker is MBL 101E and the dealer told me that BACCH-SP will not work with MBL 101E.

The best analogy is like the difference between 2D and 3D movie. However, some of the best pictures are from 2D if done it right and it can be very real. And some 3D movies are really poor. I think it is the same for BACCH-SP. It really works well with classical music but to me not so well on some recordings of rock music. Fortunately, you can turn the filter on and off with just a button in iPad.

I want some more demonstration of BACCH-SP with different speakers. I did think the sound from Sander are not very good. But be prepared, it maybe really the future of listening music.

 

[Fitzcaraldo215]

Blumlein is as old as the hills, and theoretically "perfect", but... It has had its chance to succeed and has not due to a number of factors. About the only label of note still using it is Water Lily, if they even still exist. So, you are dealing with old mostly analog recordings, and not too many of those comparatively.

Then there is the issue of speaker or listening position for plus/minus 45 degrees, which is incompatible with all other stereo recordings, and you have a tiny sweet spot. Also, engineers tend to prefer omni mikes because they are less colored than the directional mikes Blumlein requires. Blumlein had its chance, but I doubt it is going to make a comeback.

 

[Atmasphere]

The purpose of the BACCH-SP is to allow the spatial relationships on the recording to be fully heard. Putting speakers in a room causes you to hear many distortions--principally the crosstalk between the speakers but also reflections from the walls, etc.--that the theoretically perfect stereo system would not cause you to hear. By mostly eliminating the crosstalk between the speakers and mitigating some other distortions with very precise EQ, the BACCH-SP represents an advance toward the theoretical ideal.

Here is something interesting: What you describe here is what the Audiokinesis Zephrin loudspeaker also does- but by adding some crosstalk which has a late delay to it. In this way it takes advantage of the human ear/brain perceptual rules to increase the perception of image and depth. But the way you describe this I get the impression that the designer, while perhaps a good rocket scientist, really does not understand human hearing perceptual rules all that well and has come up with a fix for something that does not exist. IOW, crosstalk between speakers is known to increase spatial perception, not decrease it.

However we have a simple means of sorting out if this system is real: if the market adopts it, then it is likely so.

Blumlein is as old as the hills, and theoretically "perfect", but... It has had its chance to succeed and has not due to a number of factors. About the only label of note still using it is Water Lily, if they even still exist. So, you are dealing with old mostly analog recordings, and not too many of those comparatively.

Then there is the issue of speaker or listening position for plus/minus 45 degrees, which is incompatible with all other stereo recordings, and you have a tiny sweet spot. Also, engineers tend to prefer omni mikes because they are less colored than the directional mikes Blumlein requires. Blumlein had its chance, but I doubt it is going to make a comeback.

If you know what 2-channel stereo is, Alan Blumlein created a lot of that; he was fundamental to the invention of stereo. All stereo recordings are based on the Blumlein patent, which did not rely on the figure-8 microphones (although he did favor that technique), in his patent Blumlein was also using omnis as figure-8s had not yet been invented. So we might be having a semantic issue; when I talk about Blumlein I am talking about the many processes that Alan Blumlein created.