Tag Archives: Digital-to-analog converters

I’m pretty sure that Edgar Choueiri is the smartest person I’ve ever met. He’s a tenured professor of Applied Physics at Princeton who has, for more than 25 years, run the University’s NASA-funded Electric Propulsion and Plasma Dynamic Laboratory. Just about every interview or article about Choueiri that I’ve encountered delights in noting that he’s an “actual rocket scientist,” or something to that effect. He’s fluent in four languages, an expert in ancient Greek vases, an amateur magician, and a practitioner of Zajal, a centuries-old Arabic form of improvised chanted debate. Edgar Choueiri is also a self-described audio nerd, which distinguishes him from all the other Ivy League astrophysicists I know.

If you visit Dr. Choueiri’s EPPDyL facility, after admiring the two enormous rocket engines set up in a large, high ceilinged, garage-like space, you can exit via a discreet side door to enter Choueiri’s other academic realm, Princeton’s 3D Audio and Applied Acoustics (3D3A) laboratory. The Lebanese-born Choueiri has been obsessed with “spatial audio” since he was a teenager, assembling a system to play quadraphonic LPs at the age of 14. (He owned exactly one four-channel record, an Art Blakey title that was clearly “in heavy rotation.”) Early on, Dr. Choueiri concluded that audio theorists and engineers had made less progress with understanding the spatial aspects of sound reproduction than they had with other sonic parameters. For years, he’s been a tireless investigator in the field, publishing numerous scholarly papers and creating both hardware and software with broad utility. In 2014, Choueiri founded Theoretica Applied Physics to manufacture and market audiophile-caliber components incorporating his patented innovations.

There’s a problem, Choueiri and many others maintain, with the way that stereo recordings have been played back for the last 70 years or so. “If you go out in the forest and you hear a bird singing, it’s not because there are two birds singing,” Choueiri explained with his characteristic intensity. “There’s one bird singing.” Stereo only creates the illusion of localized sound by manufacturing a phantom image “and your brain doesn’t believe it.” In life, a sound is precisely localized because of a slight difference in the arrival time at the right and left ears, as well as slight differences in amplitude and tonality that are attributable to the physical presence of the listener’s head and the shape of his or her ears. With reproduced sounds emanating from two loudspeakers, these relationships are considerably degraded, especially if the listening environment introduces reflections. Each ear isn’t hearing what it’s supposed to—inter-aural crosstalk is spoiling the party.

In the early1960s, two Bell Labs scientists, Bishnu Atal and Manfred Schroeder, invented the signal processing technique of crosstalk cancellation (XTC) to address the matter. Over the next several decades, acoustic researchers and engineers worked to develop the methodology to the point where an XTC algorithm made it into a number of commercial products, including gear from Polk Audio and Bob Carver. But there were issues with early XTC efforts. Too small a degree of XTC can result in a kind of spaciousness that may be pleasant but really doesn’t advance the cause of realism. In addition, it was necessary for a listener to sit perfectly still, not moving his or her head at all, for the effect to be consistently experienced. Finally, early XTC algorithms caused gross timbral colorations. “Tonally, it was a mess,” Choueiri observed. “A piano sounded like a xylophone.” The crosstalk-cancellation filter that Edgar Choueiri has invented is called BACCH, for “Band-Assembled Crosstalk Cancellation Hierarchy.” It should surprise no one that the professor’s favorite piece of music is the B minor Mass, by one Johann Sebastian Bach.

The BACCH filter aims to solve the major well-known shortcomings of previous XTC schemes. Choueiri developed a sophisticated head-tracking mechanism that considerably enlarges the “sweet spot” for the primary listener and obviates the need to sit in your chair as though rigor mortis has set in. More critically, the BACCH filter doesn’t introduce any coloration to the signal. How is it done? At the most basic level, Choueiri found a way to shift XTC processing from the amplitude domain to the more “subliminal” phase domain, a manipulation of the signal that the brain is less likely to notice. The BACCH filter is the central feature of Theoretica’s commercial audiophile products.

There are three “stereo purifiers,” all with the same BACCH XTC filter. The $54,000 Grand BACCH-SP sports a substantially larger and heavier enclosure—often the most expensive “part” in an audio component—as well as 6-channel DAC and ADC cards, soup-to-nuts connectivity, and a slew of hardware and software features that are standard. It’s the BACCH-SP model for audiophiles who require all the connectivity options, including AES/EBU. The other two models are the $23,800 BACCH-SP adio considered here, which also includes 6-channel digital converters and analog inputs/outputs, while the BACCH-SP dio, at $19,800 lacks the converters and analog connectivity. (Just so you know: adio = analog and digital inputs and outputs; dio = digital inputs and outputs.)

The sculptural aluminum chassis, available in either a silver or black matte finish, is manufactured for Theoretica by MSB Technology in Watsonville, CA. In front is only a single power button with a status indicator above it that glows amber to let you know you’re up and running. To the rear, in addition to an IEC power cord receptacle, is a generous selection of connectors. On the digital side are USB, SPDIF (RCA), TosLink, and inputs/outputs to utilize an external clock. There are balanced analog ins and outs (XLR, RCA, and TRS), a headphone jack—it’s perhaps a little peculiar to have this in back but it does make for a cleaner look anteriorly—an Ethernet port, inputs for the supplied in-ear measurement microphone apparatus, and USB inputs for both an IR camera and a webcam, one of which is needed to utilize BACCH-SP’s advanced head-tracking capabilities.

It’s harder to provide information about what’s inside the box, and Choueiri really doesn’t want to dwell on it; he’d much rather get into a discussion about the processor’s unique tech and functionality. To run BACCH’s proprietary algorithms and convolution engines is a powerful multicore CPU with 64-bit audio processing; the linear power supply, like the chassis, is sourced from MSB. All BACCH-SP models come with a dedicated iPad with the very user-friendly GUI loaded in. Dr. Choueiri won’t say who makes the ADC and DAC chips, only that they have state-of-the-art jitter control, are customized by Theoretica for implementation in the processor, are of the sigma-delta type, and operate at a resolution of up to 24-bit/192kHz. Included with the BACCH-SP adio are the tiny in-ear microphones, built in Choueiri’s lab, needed to create an XTC filter for a unique listener, as described below. Finally, for the head-tracking function, the purchaser gets a webcam that’s mounted in front of the listening position.

Setting up the BACCH-SP is remarkably quick and easy. With the iPad in his or her lap, the user accesses the “Make Filter” screen—one of just three used with any regularity—and indicates whether the filter is being created for listening through speakers or headphones, and whether head tracking is desired. The miniature microphones are fitted with soft rubber earplugs (three sizes are provided) and the user inserts them into his or her ear canals, taking care to assure a tight fit but not letting the microphones touch the walls of the outer ear. The listener picks one of seven “bins” where the filter can be stored; other listeners can keep theirs in another. Then it’s ready, aim, fire. The user presses a flashing green button and a soothing, reassuring voice provides further instructions. The BACCH-SP adio will then emit a total of six full-frequency sweeps—one through each loudspeaker with the listener facing straight ahead, two sweeps with the person leaning a foot or so to the left, and then two more while leaning to the right. That’s it. It takes all of about two minutes to make the XTC filter and establish head tracking. You’ll need to do it again if you change the prime listening location, reposition the speakers, get new ones, or make significant changes to the room, such as a new piece of furniture or window treatments. The lateral extension of the sweet spot is adjustable, but if it’s too wide and there’s someone sitting next to you, no one gets the XTC experience—the algorithm sees too many heads and defaults to the center position. However, the BACCH filter works well for several yards immediately behind the prime listening position, which is why, at audio shows, Dr. Choueiri will place chairs in a line leading backward from the sweet spot.

There’s a “Part 2” to the set-up story. As noted earlier, room reflections are the sworn enemies of crosstalk cancellation, and most domestic listening environments are afflicted with them to some degree. There are three primary ways to address this reality. First, the radiation pattern of your loudspeaker matters a lot. Directional speakers—dipoles, electrostatic designs, horns with waveguides—have a leg up. Less directional speakers, such as my Magico M2s, are at a relative disadvantage. Second, one can sit in the nearfield. Third, there’s room treatment. Professor Choueiri visited my room twice. On the first occasion, he incorporated the BACCH-SP into my system, showed me how to use it, assured the device was functioning as it should—and left the speakers and room alone. On his subsequent visit, though, Choueiri arrived with a carload of acoustic materials and deployed 11 pieces of absorptive foam and two large stand-mounted RealTraps, their disposition guided by measurements obtained with the BACCH-SP using the microphones inserted into my ears, while I was sitting in the listening position. Additionally, the M2s were moved a foot nearer to the listening position and about two feet closer together, adjustments in the direction of a nearfield orientation. These changes were implemented to maximize the results of the multichannel emulation experiment (by minimizing the effects of reflected sound), described in the sidebar below. They were not needed to reap the benefits of the BACCH filter with routine listening; I removed the acoustic treatments and returned the Magicos to their usual location after the multichannel investigations.

I used the BACCH-SP adio as a processor between my 432 EVO Aeon server and Ideon Absolute Epsilon DAC, and as a stand-alone digital front-end, employing the adio’s internal player and DAC. I played music stored on my Synology NAS and streamed via Tidal and Qobuz. I also played silver discs with a Sony X1100ES as the transport. The BACCH-SP and the Ideon sent their analog output to a Pass Labs XP-22 linestage; power amplifiers were the Tidal Ferios monoblocks that have recently rocked my audio world, driving Magico M2 loudspeakers.

Listening to music played with the BACCH-SP adio for six weeks can only be described as revelatory, in the sense that I heard a domestic two-channel system—mine—do things I’d never heard before in any setting. Two observations should be made at the start. First, as promised, the BACCH XTC filter introduced no colorations or timbral distortions to the reproduced sound. Instantaneous comparisons of filter/no filter are readily accomplished with the BACCH app on the iPad; it’s easy to toggle back and forth between the bin your filter is in and “Bypass.” The crosstalk-cancellation process is utterly transparent. Secondly, whoever did design the DAC (and ADC) for Theoretica did a helluva job. I heard no meaningful difference in the overall sound of my system when using the $47k Ideon or the BACCH-SP. Getting the model with the DAC (and ADC) requires an additional expenditure of $4000. Sounds like a bargain to me.

What does the BACCH-SP XTC filter bring to the table, sonically? The processor renders several aspects of spatiality very effectively, some of which I’ve experienced only rarely in the past with the most elite systems, if I’ve heard them at all. One is envelopment. The sonic image moved out in front of the two speakers and wrapped around the sides to end up well out into the room, outside the lateral boundaries of the Magicos. On a Chesky Records binaural recording featuring trombonist Wycliff Gordon (Dreams of New Orleans), Gordon is localized at the 10 o’clock position, forward from the plane of the loudspeakers. His improvisatory genius is believably isolated and exposed—your brain tells you that’s where he is standing as the featured performer, not where some mixing engineer put him in post-production. Music played back with the filter engaged often seems subjectively louder, perhaps because of this heightened sense of immediacy.

Proximity/depth is an attribute of spatiality that Dr. Choueiri uses to seduce audiophiles when he demonstrates BACCH at shows. On “Phrases,” from another Chesky release (Dr. Chesky’s Sensational, Fantastic, and Simply Amazing Binaural Sound Show), several voices seem to be whispering just inches away from your ears. With musical content, this can translate into a kind of layered depth that’s more subtly defined than with traditional stereo. Another spatial metric is reverb. Manual de Falla’s ballet The Three-Cornered Hat, as memorably recorded by Bert Whyte for Everest in 1960, features, in the “Introduction,” a series of aggressive sounds—castanets, stamping feet, testosterone-fueled yells, timpani thwacks, a caterwauling soprano—all illuminating the space of Walthamstow Assembly Hall in London. With BACCH, the reverberant tail of all these sounds is heard as being attached to the initial impulse but doesn’t obscure it. Without the filter, the recording is electrifying, even atmospheric; once you’ve heard playback via BACCH, however, it can be hard to go back. Without BACCH, something that can only be described as “a sense of occasion” is missing.

Envelopment and proximity are attention grabbing at audio shows but can be dismissed by skeptics as gimmicky; reverb can be taken for granted. A fourth spatial characteristic of the sound created by a BACCH filter, the one that impressed me the most, is what Choueiri has called spatial extent and resolution. Extent, he explains, “is the perception that the sound occupies a three-dimensional volume, like a hologram,” while resolution, in this context, “is the ability to discern detail and structure within the extent.” Listening to my favorite orchestral test track, the opening Allegretto of the Shostakovich Symphony No.15, as performed by Bernard Haitink and the Royal Concertgebouw Orchestra, I attended closely to the sequential woodwind solos near the beginning of the movement. It’s not only clear that the bassoon is a larger instrument than the flute, and that the former is seated a row behind the latter—you can get that kind of information from traditional stereo—but also that these correctly scaled and localized aural images interact to represent a continuous acoustic environment where the musicians are breathing the same air but still own a unique three-dimensional space of their own. It’s like the difference between the pop-up-book kind of depth and dimensionality you get from most 3D movies and the far more effortless perception of space you experience in life.

This can be extremely gratifying with no small number of 50-year-old rock/pop albums, supporting Choueiri’s contention that BACCH works not just with “dummy-head” recordings but also with many—most—conventional stereo albums of recent and not-so-recent vintage, whatever the recording methodology. On “You Don’t Have to Cry” from CSN’s epochal first album, every one of the four or five (at least) acoustic guitars was played by Stephen Stills. Even with the best remasterings of the album that I’ve heard, those guitars come off as a busy, almost Baroque counterpoint to the triadic vocal harmonies they support, like one gigantic plucked/strummed instrument. With the BACCH filter engaged, each guitar is thrown into bold relief, staking out specific real estate and presenting a fleshed-out sonic image. It becomes obvious that Stills arrived at the recording session—the first ever for the three musicians—with a clear idea of how his ingeniously interlocking guitar riffs would come together as more than the sum of their parts. The song is about Stills attempting to convince Judy Collins, his inamorata at the time, to move from New York to California. (Sensibly, she declined, fearing the inevitability of a slide into substance abuse.) With BACCH, the individualized guitars become many more voices talking past each other, symptomatic of a doomed relationship.

Choueiri allows that there’s a hierarchy of recordings, in terms of how well they will fare with BACCH. Binaural recordings will be most obviously suitable, followed by “purist” recordings of jazz and classical. Generic recordings of acoustic music are next, followed by “concocted” rock and pop studio efforts. But there are unexpected pleasures to be had even with the popiest of pop recordings. Check out Elton John’s “Nikita”—subtly, but consistently, there’s an off-the-beat drum sound that’s hanging out at two o’clock. It has to make you smile.

It’s only natural to focus on the processor’s clarification of individual instrumental lines and the physical disposition of the performers, but I shouldn’t neglect to point out that crosstalk cancellation, as executed by the BACCH filter, can elucidate harmonic detail, as well. The characteristically dense orchestral textures of composers like Brahms, Richard Strauss, or Messiaen can seem murky as represented on recordings, which isn’t the case in the concert hall. BACCH can improve upon that artifact of stereo playback to a significant degree. It’s true, as well, with complex vocal arrangements in popular music genres. Compare Donald Fagen’s extravagant harmonies on the title track of Morph the Cat with the filter on and off, and you’ll know what I mean; prepare to enjoy Queen’s most operatic moments to the fullest—the first 45 seconds of “Bohemian Rhapsody” makes the point nicely.

An option available on both the adio and dio models (and standard on the Grand) is Theoretica’s BACCH-hp filter. This technology can make headphone listening much more appealing to audiophiles who, like me, don’t care much for “personal stereo.” Two aspects of the filter are notable. First, the image presented to the headphone wearer—Theoretica recommends open-back designs—is much more externalized than with typical ’phones. For the uninitiated, the experience can be quite discombobulating: Sitting in the sweet spot with HiFiMan HE400se headphones on, I was absolutely convinced that the stereo sound was coming from the Magicos eight feet in front of me. I had to remove the headphones to confirm that the speakers were indeed silent. It was uncanny. (Sorry.) The second innovation involves head tracking. With typical headphones, if the listener turns his or her head, the musical image moves along with the change in orientation—not exactly a realistic effect. When setting up a BACCH-hp filter, the user is instructed to rotate his or her head, as opposed to leaning right or left as directed during the calibration of BACCH-SP for loudspeakers. This “rotational head tracking” anchors the 3D image in space, even when the listener looks right or left. Theoretica ships all its players equipped with the BACCH-hp filter and gives a new customer a few weeks to decide if, for $3k, he or she wants it. If not, Theoretica Applied Physics can turn off that feature remotely.

Any discussion regarding spatiality in sound reproduction needs to touch, of course, on multichannel (see sidebar). Though Disney’s Fantasia featured surround sound in 1940, commercial efforts began in earnest with Quadraphonic in the 1960s and 70s, fizzled out, and then got a shot in the arm with the rise of home theater and the subsequent introduction of the SACD, DVD-A, and Blu-ray Disc formats. Currently, we’re seeing further enthusiasm for what’s now often referred to as “immersive sound,” with more channels and more effective encoding schemes—Dolby Atmos, among others.

To be sure, there are some aspects of spatial sound, envelopment in particular, that traditional, discreet multichannel does better than BACCH, at least for the time being. But that doesn’t diminish my enthusiasm for Theoretica’s product. The BACCH-SP devices are intended for two-channel aficionados who want to extract the best possible experience from a stereo setup: There are literally millions of stereo recordings out there that have the potential for a completely unanticipated improvement in sound quality. The price tag of more than $20,000 for the adio model may seem less daunting when you consider that the device contains a world-class DAC and can serve as a transparent full-function preamplifier. I even connected a phono-    stage to one of the BACCH-SP adio’s analog inputs and heard some favorite LPs get the XTC treatment with excellent results. Watch for the BACCH-SP filter to be licensed by other high-end manufacturers—it’s already happening.

The composer, jazz pianist, and HDtracks co-founder David Chesky knows Edgar Choueiri well, the Princeton professor having been involved with many of Chesky’s own binaural recording projects. “We have been stuck in this 60-degree stereo triangle,” Chesky told me. “Think of surround as a hula-hoop. It’s a big one-dimensional circle around you. BACCH is more like concentric circles that get bigger and bigger to emulate how we hear in real life.” The BACCH-SP adio is a logical consideration for two-channel audiophiles looking to correct a fundamental deficiency of stereo playback. As of this moment, BACCH doesn’t supplant multichannel. The two approaches have different things to contribute in terms of the ongoing quest for spatial realism. But if there is a way for an XTC system to better emulate all the attributes of an existing multichannel mix, I’m sure Dr. Choueiri will figure it out. For him, at least, it’s not rocket science.

Specs & Pricing

Type: adio model with preamp/processor and DAC, crosstalk-cancellation filter, and head-tracking capability
Analog inputs/outputs: RCA, XLR, TRS (tip, ring, and sleeve)
Digital inputs/outputs: USB, SPDIF RCA, SPDIF TosLink, word clock (BNC)
Included: BACCH-BM in-ear binaural microphones, iPad, webcam
Dimensions: 17½” x 3¼” x 13½”
Weight: 22 lbs.
Price: $23,800 base price. As configured (with BACCH-hp software and 3D Audio Analysis Toolkit) $27,800

THEORETICA APPLIED PHYSICS
417 Alexander St.
Princeton, NJ 08540
theoretica.us

The post Theoretica Applied Physics BACCH-SP adio Stereo Purifier appeared first on The Absolute Sound.