Theoretica Applied Physics BACCH4Mac Stereo Purifier Review

BACCH processing previously has been done via a special box and software that is quite expensive. Andy Quint reviewed this system here a year or so ago. Recently, Theoretica released a version that runs on Apple Mac computers and gets the price point down from circa $20k to under $2k. There are various versions of BACCH4Mac, starting at $980 plus the cost of a Mac, assuming you don’t have an M2 Mac lying around unused. So, about $1600 for the most basic version. The fancier versions go up to about $8700 plus the Mac for the Pro setup, but you probably don’t need the recording capabilities or the headphone processing of maximum supreme BACCH4Mac, so figure $6700 plus the Mac. This is called the Audiophile version, which offers head movement tracking and custom compensation for the exact shape of your head.

We reviewed the BACCH4Mac Pro version, but it does exactly what the Audiophile version does when used in a stereo system. The software is somewhat complex to set up, but Theoretica does a call with you and runs your Mac remotely so that they are both showing you how to do setup and actually doing the setup. This takes perhaps an hour, and you are good to go. After that, you really don’t have to touch the software unless you change your stereo setup.

The BACCH4Mac was used in my reference system as follows:

• Ethernet from router to MacMini (M2)
• Roon Core running on MacMini
• Roon Core streaming Qobuz and Tidal to BACCH4Mac on the same MacMini
• Head-tracking camera connected via USB to MacMini
• Babyface Pro mic preamp enabling low-level software license used by BACCH4Mac
• MacMini output via USB to dCS Lina DAC (with Master Clock)
• dCS Lina output to Audio Research Ref 6 mk2 preamp
• Ref 6 to PS Audio BHK 600 mono amps
• BHK 600 amps to Magico A5 speakers

I ran other digital sources straight into the dCS DAC. Analog sources went straight to the Audio Research preamp. This allowed me to compare various signals to those going through BACCH4Mac without ever touching the MacMini or BACCH. BACCH4Mac can do A/D conversion if you want to use it with analog sources.

Okay, But What Problem Are We Trying To Solve Here?

We have said, and it bears repeating, that there are at least three great unsolved issues with stereo music reproduction in the home:

• The problem of bass in real rooms
• The problem of visual imagery
• The problem of spatial presentation

BACCH4Mac is built to address the latter of these. That is, some people, your humble reporter included, observe that the soundspace and instrument placement and rendering of instruments in conventional stereo is not fully realistic. A common problem is that performers seem to be paying from within the speakers at times. I also like Siegfried Linkwitz’ version of the problem, which is that small performers seem to be playing on a stage arrayed between the speakers. Audiophiles, myself included, work hard to make this stage as wide and as deep as is realistic, and then get performers placed on it correctly. Audio companies and recording engineers work hard to enable this, but generally within a framework that accepts these constrained stage limitations. BACCH asks “why is this limited soundspace happening in the first place and can we address it?”

Okay, fasten your seatbelts because this is the part of the presentation where we have to unseat some common assumptions. At least my experience was that some unseating is needed.

The first assumption that I find many audiophiles have is that stereo is designed to allow placement of performers anywhere between the speakers. That, often, is what we think stereo means. So, if a performer is on the left side of the stage, he or she would be presented from the left speaker. If the performer is in the middle of the stage, he or she would be presented in the middle. If the performer is center-right, he or she would be presented between the middle and the right speaker.

But if you go back to the original Blumlein patent for stereo from 1931, this wasn’t the idea. Or, said differently, this ‘between the speakers’ idea is wrong. Instead, stereo is intended to be able to present the performers where they were during the performance, unrestricted by speaker location.

To further understand what the idea was, you may have to give up another idea: many people think the speakers are delivering a version of the instrument sounds that were recorded and presenting them into the room in a simulation of what happened in the concert hall or the club. You then hear something that is roughly like what you would have heard live. The speakers set up a virtual performers and the reflections in the room create a sense of the space.

But this is not what happens, or at least what should happen. What actually happens is that the recording consists of two data streams, one for the left channel and one for the right channel. These data streams have all the information about performer placement encoded in them. That is, if a performer were to the left of the mics used for the left channel, the data about that leftward placement is on the recording. This data is sent to your ears from the speakers, and here’s the kicker: your ear and brain have the ability to decode the data and place the performer where he or she was when the recording was made. Which means that a performer who was to the left of the left mic can be placed so that you will perceive it being to the left of the left speaker.

Yup, your speakers are data conduits not performer simulators. Let that sink in for a minute.

Now, I still haven’t really explained where the problem lies. So, here we go. My apologies if your head hurts.

(It is important to note that the explanation below is a greatly simplified summary of how this works. As a simplification, it is intended to give you a feel for why normal stereo doesn’t work right and very roughly how it can be fixed. If you come away with that sense that there is an addressable problem, then I urge you to let the proof be in the pudding and attend to the sonic results of BACCH which I describe next. And listen for yourself. But, if you are the kind of person who wants to follow the logic in detail, step-by-step, this won’t get you there. There are 52 pieces of intellectual property in BACCH and Blumlein made 70 key points in his stereo patent and there is more that has been learned about stereo in between. I’ve reduced this to three steps and inevitably that doesn’t get you anything like all the detail to prove that BACCH makes sense. You’re either an empiricist at heart or you’re not, and if you’re not what follows won’t help.)

Let’s start with a live performance. To keep this as simple as possible, I’m going to concentrate on what happens to your left ear and to the left microphone. But, it is very important to say, the right ear and right mic are logically just like this. I’m also going to simplify by assuming just two performers and just two mics. The logic would be the same with more performers and hall reflections and more mics, but the algebra gets messy.

Okay, take a look at this diagram. What your left ear hears in a live performance is shown here. Your left ear hears the direct sound from the left performer. Your left ear also hears the sound from the right performer but modified by the distance and time delay and frequency shaping caused by your head. This is called the head-related transfer function, but I’ve simplified this by calling what the left ear hears from the right performer R(ModH) for “Right Modified by Head”.  I’ll come back to this, but ideally your stereo would present your left ear with L+R(ModH). You will soon see that it doesn’t.

Now, consider what happens when we make a recording of these two performers. Instead of your head, I’ve shown two mics in a Blumlein arrangement. The left mic, which will feed the left channel, records the direct sound from the left performer. It also records the sound from the right performer, modified by the distance and time delay and off-axis pickup pattern of the left mic. So, in my algebraic representation, the left channel of the recording is L+R(ModM) or left plus right modified by the mic. Now this isn’t exactly what your left ear heard, but it is similar in form.

Then consider what you hear on your stereo. Again, we will just consider what your left ear hears to keep it simple. From the left speaker, your left ear hears L+R(ModM). So far so good. That is pretty similar to the data stream your left ear heard in the concert hall, which is L+R(ModH). But, your left ear also hears sound from the right speaker, which is R+L(ModM). Now hang with me on thinking of these as data streams not virtual performers. When you think of it that way, you realize that with standard stereo we have a huge error term. Everything from the right speaker that your left ear hears shouldn’t be there! You are hearing R+L(ModM) in the left ear and you should not hear any of that term in the left ear in order for the left ear to hear something like what it would hear in the concert venue.

Now, to understand BACCH, I want you to consider that standard stereo presents a massive distortion signal to each ear. That distortion signal is R+L(ModM) for the left ear. Assuming that the right speaker signal at the left ear is roughly the same magnitude as the left speaker signal, standard stereo has something like 50% distortion. Woah! Now this isn’t a-musical distortion, it isn’t for example harmonic distortion or noise, so that number overplays how audibly distracting this error term is. But, the error term is big. And it is spatially distorting.

Getting back to BACCH processing, the BACCH software removes the error term. They call it crosstalk cancellation, but it isn’t the internal to equipment crosstalk that they’re talking about. It is basically an additive signal that is imposed by stereo reproduction that adds a second erroneous layer of music on top of what happened in the hall or club. This crosstalk removal is “all” BACCH does. So maybe the magnitude of the change is more understandable if I say BACCH processing removes a 50% distortion.

Now, removing this amount of error or distortion in practice without damaging the music along the way is no simple task. So, when I discuss the sound quality of BACCH I will address how well it does its work without extraneous damage. Stay tuned.

How Does BACCH Processing Sound?

BACCH makes a dramatic difference to the sound you hear. That difference varies from recording to recording.

The recording-to-recording observation brings up important point number 1. BACCH processing is not adding something to the signal in an attempt to simulate what you would have heard in the concert hall or club or recording studio. BACCH is subtracting a signal that shouldn’t be there to properly render the performers in space.

Let me give you some examples of how BACCH processing sounds.

With the Mozart string quintets, K515 and 516 with the Eben Quartet in high res, you first notice that the quintet is spread out in a very clear semi-circle, and the semi-circle extends very slightly beyond the width of the speakers. Then you notice that each instrument seems to stand out in space so it sounds like the players are actually occupying not just the semicircular positions that I described before but that each player seems to have an instrument that’s got a whole spatial property to itself.  As it would in real life if it were actually in your room resonating. In addition to the semicircular array of the instruments, the players on my reference system, particularly second violin and viola, seem to be behind the front wall. The stage depth with BACCH is deeper than just working in the space behind the speakers.

Next up I listened to The Freewheeling Bob Dylan from 1963 in high res. This is a really simple recording: it’s Bob, his guitar and his harmonica. With BACCH, you hear Bob in the center where you would kind of expect him if you think about how most stage performances are arranged, and then the recording engineers put his guitar on the right and the harmonica on the left. That isn’t how it would sound live, but this is a studio recording and you hear how the engineers placed things so that it sounds a bit like a trio. Apart from the deviation between studio and live, what’s impressive is the guitar occupies a space that extends outside of the right speaker and feels appropriately sized for the dreadnought Gibson guitar that Bob played in this period. This album does raise a persistent BACCH conundrum which is: do you want to hear musicians where the recording engineers placed them? You’ll have this issue in normal stereo of course, just not so clearly presented.

With the Esbjorn Svensson Trio, on the jazz album From Gagarin’s Point of View, the mix seems to me to be just about perfect with the performers arrayed on a stage in front of you with a real sense of depth. The tonality of the instruments is beautifully rendered, but the part that the BACCH processor adds if you will, is the sense that the players are playing in a large three-dimensional space. The reverb off the rear wall and the sidewalls and the front wall just seems so much more natural and like they were actually playing in a club, that it’s borderline transformative. The reason it feels transformative is that this added realism subconsciously helps you suspend disbelief. “Realism,” it turns out, isn’t an analytical, nerdy thing alone, it is a critical part of engaging with the music.

I tried the Mahler Symphony No. 1, this time with the Minnesota Orchestra. This is a Hi-Res recording made in 2019. Here I had a similar feeling that BACCH processing enhanced the sense of the orchestra being in a large three-dimensional space. instruments that are farther toward the back of the hall were presented properly, with horns and timpani appearing toward the back of the hall. The reverberant field is so well captured that the sense of the musical reflections floating above the orchestra is a little bit stronger than what I’m accustomed to with stereo, but it is very much like what you hear when you’re actually in the concert hall.

Now let’s talk about Pink Floyd’s Dark Side of the Moon. This is a recording where I’m pretty confident the band wanted a surround sound kind of effect rather a simulation of what it would be like to hear the band playing live on a stage 200 feet away from you. The famous first track with the heartbeat is called “Speak to Me”. Toward the end of the track there are some voices and helicopter effects and chimes, and these are placed way outside of the normal stereo stage area. In fact, these are located slightly behind the listener in some cases. I mean that you hear the sounds as if they were coming from a speaker set behind you and to the left or right. But there is no speaker there. As I said, the data is locating the sound there, and is not muddled via BACCH, so you hear it where the recording team presumably wanted it. The helicopter and voices also move around laterally in a clear way as if they were moving around your room.

Then we move onto the track “Breathe”, and the listener is placed in the center of the band. David Gilmour‘s guitar is to the left and slightly behind the listener. Richard Wright’s organ is to the right rear of the listener and Roger Waters and Nick Mason are presented on the traditional stereo stage in front of you. I do feel that the instruments move around a little bit and that again seems to have been intentional. It’s very much a surround sound, listener in the center of the band, psychedelic kind of thing.

Listening to Radiohead on the In Rainbows album, I was again struck by the ability of BACCH processing to take each instrument and give it its own place in the mix without blurring it together with other instruments. So, I can honestly say I haven’t heard this album sound as good on my reference system as it does with BACCH. Now we also need to consider that this appears to be an album where the engineering places some of the instruments outside of the normal stereo stage. I assume that’s what Radiohead wanted. You might find that slightly unnerving, although for this kind of music, it seems somehow appropriate. And it was very engaging. The approach is a little bit like the semi-psychedelic stuff from Pink Floyd, but the recorded effects were more subtly presented here than on Dark Side of the Moon.

I do want to describe an addressable downside of BACCH processing. On Art Blakey and the Jazz Messenger’s album Buhaina’s Delight, the players are arranged in a 270° arc around the listening position that may be effectively where the microphones were. However, I have to say it’s a little bit odd to hear Art Blakey playing in a position slightly behind where I’m sitting, and on the right side. There is a horn off my off my left side. The other players are arrayed as you would typically hear on stereo between the speakers. Not what you would hear in a club. This studio recording is from 1961 and I just don’t think stereo recording in the studio was evolved to the point where the engineers were trying to do this. I think the engineers wanted Blakey right front so they pan-potted him hard right and the data on the recording has this placed at the right rear. I got similar results with Dave Brubeck’s Time Out, from 1959. But Bill Frissell’s Valentine, a jazz trio studio recording from 2020, sounds phenomenal in the sense that the performers are where you would expect them.

You can set up BACCH processing with choices of different levels of crosstalk cancellation to address this issue. With 25% rather than 100% cancellation, the Art Blakey recording sounded more normal and I think correct.

The fact that recordings can present performers in a 360-degree arc with BACCH processing doesn’t mean that all recordings will be spread out in that arc. BACCH isn’t imposing a pre-determined “envelopment factor” on the recording. It that means that 360 degrees is the maximum placement freedom that the engineers can have in choosing where the recording places sound. If you’re on full crosstalk cancellation and the engineer has placed somebody far left and behind you, essentially -135° from dead center or +135° from dead center, you’ll hear them there. But if the engineer placed the musician in the middle between the speakers, as with the Bob Dylan recording, the performer is going to be in between the two speakers in the center just like you’re used to from time immemorial. So, with BACCH processing, the stage width will vary from recording to recording, and dramatically so.

Now there is a reasonable argument that the recording engineers probably didn’t listen with BACCH when mixing and mastering. So, they expected playback via normal, cross-talk laden stereo. So, BACCH represents a distortion of what they wanted if you assume they were able to get what they wanted. I will counter that with the observation that many modern recordings seem to be made in a way that manages to take studio work and live work and present it spatially in an okay rendition of space that become significantly more realistic with BACCH. That is, without BACCH, a good recording is rendered within the limits of normal stereo in a good spatial manner. But the techniques used to present the performers spatially on limited stereo systems seem to encode the performers in a way that has better data than you normally hear. Until you use BACCH.

An analogy may help. It is as if the engineers know that bass below 50hz is important. So, they develop techniques to get flat, accurate bass from 25-50hz, even though their studio monitors roll off below 50 hz. Then, when you play these recordings on a system flat to 25hz, it sounds great and natural. Listening on rolled-off monitors wouldn’t support this, so they develop rules to get around the limits of the system. The engineers are smart enough to make the recording accurate, even though their monitors don’t let them hear it.

I do think mixing and mastering with BACCH would be even better, of course. But I am suggesting that, in a lesser of two evils world, current recordings via BACCH often (>60% of the time) sound more spatially convincing than without BACCH.

All this talk about arcs and performer placement may do an injustice to one of the big benefits of BACCH. To help you imagine this other element, I would say BACCH seems to give each instrument not just its place on the stage, but it’s own sense of 3D air and size. It’s almost as if each instrument gets its own speaker and that speaker can devote all its energy to perfectly representing the sound of that one instrument. This idea that each instrument has its own channel gives you a sense of the purity and the clarity of harmonics and instrumental size and detail that BACCH delivers. This could be summarized as a sense of higher resolution. I want to keep emphasizing that this clarity or resolution isn’t spectral byproduct, it’s a spatial byproduct. You can better hear into what’s going on in the instrument.

And you get more suspension of disbelief. The more that the spatial qualities don’t appear locked into an artificial construct (e.g. the speakers or the tiny stage) the more you listen to music differently and, I thought, in a more engaged way. Some of this also comes from the music seeming to be played in a real, large space. The performers are on stage as you would expect, but they are playing in a concert venue that has walls left and right and behind you. Linkwitz’ tiny stage doesn’t do this, and you know it even if years of listening has made its absence seem normal.

I also need to comment in a broader way on spectral balance. Apparently one of the hard things (there are more than 50 pieces of intellectual property here) about crosstalk cancellation is doing it without disturbing the frequency response of the music. I won’t say that the balance of my reference system was the same with and without BACCH, but it was very close to the same. I never felt the need to describe the impact of BACCH processing in frequency response terms, even though that is the way changing components often shows up in my head.

Summary

I hope it is obvious that I think BACCH processing is a significant advance in stereo reproduction. On perhaps 80% of the albums you might play, especially from the last 40 years, BACCH4Mac delivers an image that is freed up from the confines of your speakers. This gets some parts of the image off the speakers and expands stage width, depth and height in an important way. That “important way” is to make the performers seem more spatially real and to help with the suspension of disbelief that often accompanies recorded music. Performers and instruments take on a more 3-dimensional quality and their freedom for spatial placement by the recording engineers gives a 3-dimensional feel to the stage. Additionally, BACCH helps to present the ambient cues from the recording venue so that you subconsciously feel you are in a real space.

Another analogy may be useful. Conventional stereo is a bit like watching performers on a television. As we improve the many core technologies of stereo, the TV gets bigger, and the resolution gets higher, and the colors are more accurate. But you never really escape the sense that the performers are locked inside a window defined by the edges of the TV, and you often sense that the image is actually flat, even though high-res helps generate a better sense of depth. With BACCH, continuing the analogy, it is like watching movies on a hologram generator or a VR headset (without needing a headset). There isn’t a hard edge to the stage, depth is more realistic, and you feel like you are in the room with the actors who themselves are three-dimensional.

Now, recordings have been made and could continue to be made that are more in the surround-sound vein. BACCH4Mac does an impressive job with these, but I do think such recordings are musically unusual and it is important to understand that this extreme capability for placing performers is not what BACCH processing is primarily about. I want to add that I suspect that if BACCH4Mac couldn’t do accurate surround sound, it couldn’t reveal the virtual concert hall environment that I found so delightful. And, who knows, maybe if BACCH were more commonly used, some musicians would make the surround elements musically effective.

This view also should be contextualized by saying that I would consider BACCH4Mac to be a “Type 3” product, which are major game-changers addressing fundamental issues of audio. These probably only come along every few decades. That BACCH4Mac is a fundamentally new thing, and yet works with minimal side effects, is unusual and impressive.

BACCH4Mac comes to the market at a price that is quite reasonable in high-end terms. It has few downsides, if any (after all you can turn it down or off for any recording). It addresses a real issue for audio as we know it and does this for all existing stereo recordings, i.e. it doesn’t require new equipment or a change in recording technology. I do think it could encourage a refinement of recording technique, if engineers used it as part of the mix/master process, because then they would really know what data they are putting in those bits. I hope that happens, but even right now BACCH4Mac is a legitimate breakthrough.