sweetsounds
Active Member
- Joined
- Apr 24, 2019
- Messages
- 152
- Likes
- 310
So I finally broke the bank and purchased the Intro version of BACCH-DSP 4 MAC (V12.2).
Introduction
My experience is, that the interaction between room and speakers has by far the greatest impact on how I enjoy music listening. Most of my upgrades of the last 10 years went into room acoustics (without completely sacrificing the looks of the room). I also finally got good results using Acourate as a digital room correction solution to compensate for the rest.
My personal findings so far match the research of Toole/Olive, namely that a flat frequency response on axis (and an even behaviour also off-axis) makes music sound "right" and for imaging you want to either absorb first reflections or diffuse, but definitely you want to diffuse late room reflections to give a sense of space. So put diffusors at the backwall and also between the speakers. A decay time RT60 of 0.25 to 0.3 seconds works very well. Below the transition frequency use bass traps and DRC. Wide dispersion speakers yield a good compromise between tonality and sound-stage. Directional and bipolar speakers like Magnepans are difficult to set up.
What BACCH-DSP does
BACCH seems to be around since 2015, developed by Prof. Edgar Choueiri of Princeton Univ. His company Theoretica Applied Physics is now more present at HiFi shows demonstrating the effect. It is a DSP that eliminates speaker cross-talk (XTC) at the listening position by a sort of a cancellation pattern (similar to noise cancellation) while avoiding frequency artifacts due to the interference. To quote from his paper: "These two series of pulses are delayed by τc with respect to each other so that after the first positive pressure pulse arrives at the left ear, then reaches the right ear with a slightly smaller amplitude, it is cancelled there by a negative pressure pulse of equal amplitude (that was emitted a time τc earlier by the right loudspeaker), which in turn is cancelled at the left ear by a positive pressure pulse, and so on. The net result is that only the first pulse is heard and only at the left ear, i.e., with no crosstalk"
It is not a new concept, but BACCH implements it very smartly. It has special algorithms for bass and mono signal treatment and lots of calculation to minimize colorations.
BACCH DSP is available in three flavors, all based on a similar/same set of algorithms
1. Expensive: BACCH4MAC intro, at $1000. You can't customize the XTC elimination filters, but can use only a generic model based on the distance between you and your speakers (uBACCH)
2. Very Expensive: BACCH4MAC audiophile $5000/$6000 with a binaural microphone + interface to measure your system's response. It also offers head tracking by webcam to maintain the benefit when you move.
3. Ultra-Expensive: BACCH Sound Processors stand-alone equipment from $20k to $50k, which you can easily plug into your signal path.
Options 1 and 2 require an Apple Mac in the signal path to run the DSP software. This adds another $1000 and honestly reduces the usability of my system: I have not yet succeeded in making the Mac headless, so I need a monitor and mouse to configure it.
You can try option 1 for 2 weeks with full refund.
Setting it up
My current setup consists of a central music server running Logitech Media Server 8.4, a raspberry Pi endpoint running SqueezeLite on AroioOS and BruteFIR DRC. This is feeding a SMSL VMV1 DAC via coax and Classé M600 monos into Magico S5 Mk2. I always wanted to avoid the hassle and - maybe irrationally - the electrical noise of a PC in the sound-chain.
It is an easy-to-use configuration. Just switch it on and select the music with the iPeng app.
When you buy the BACCH software, a service employee (in my case it was Edgar) helps you perfectly by phone to get your system running. Support of Theoretica is unbelievable! Edgar spent more than an hour with me explaining everything.
Here is what we did: set up the Apple Mac Mini M1 16GB running SqueezePlay (you can also use Roon, HQPlayer or probably almost any other software which plays music).
You need BlackHole Loop Back software to re-route the sound on the Mac from SqueezePlay to BACCH-DSP (simply create an aggregate device in the Audio/MIDI Settings with BlackHole + your DAC as sources, set the clock source to your DAC and tickmark drift compensation for BlackHole). Call this aggregate sound device "SF4BACCH". Then in the BACCH-DSP software select SFBACCH in the Auto-Configure drop-down menu. Now select a filter bin, click "ARM". enter the distances between you and your speakers and between the speakers, and click FIRE. his is more or less it, but there are some pit-falls where Edgar helped.
Another challenge was to include my DRC filters. They need to go after the XTC filtering. BACCH-DSP supports VST/AU/VST3 plugins, so you can use a Convolver VST plugin.
Selection on the Mac is huge, but some free ones don't work. Edgar recommended Reverberate (which will cost you another $99). It is not as great as BruteFir or Accourate Convolver as it doesn't change sample rate automatically for instance.
So my signal chain now looks like this:
Server (running LMS) ==> (ethernet) ==> MacMini (running SqueezePlay, BlackHole, BACCH-DSP 12.2, Reverberate Core VST3 for the DRC FIR filters) ==> (USB) ==> DAC
More complex setups with active XOs should be possible and VST plugins like Elements open further realms for post-processing.
What is the effect
Hard to describe, it creates the impression of a 3-dimensional room in front of you, musicians and sounds eminate from a precise spots in the room rather than from between the speakers, sometimes even far outside the physical dimensions of your room, or from the side/behind you or from the top.
The effect is most evident when I switch to Bypass: the musicians are pulled with a rubberband from their positions around the speakers back to a position in the speakers' plane. The 3-dimensional room collapses.
When switching it back on, the musicians slowly move to new positions in the room, sometimes almost next to you. It takes a couple of seconds for your brain to accomodate and grasp the virtual room around you. On binaural recordings of Chesky you can have people walking up close to you and whispering. It's more than a gimmick, you forget rapidly that it's there.
The magnitude varies greatly with the musical pieces. Since cross-talk is elminated your brain is re-interpreting the spatial cues which are in the recording. If there few or none, there is also no effect and no change in the presentation.
In any case, the speakers vanish. I guess this comes from the fact, that each ear gets fed with signals coming from the respective side only. So your brain has no cues from the other ear for the speaker's position.
Which music is benefitting
Naturally, the effect is biggest with binaural recordings, but many standard recordings work as well. Best is live music like a classical orchestra, electronic music sometimes is excellent, straight studio recorded pop-rock is not benefitting much in many cases. There is a thread with a list of recordings on this site, start looking there.
Limitations of BACCH
- The system complexity is higher, you need a Mac (and I don't like Macs)
- Usability of the software is astonishingly good, but the fonts are so small, that using it in a listening environment is tedious.
- The magnitude of the spatial improvement varies a lot with the music. If studio recordings with individual mics are your thing, the benefit might be less.
- For best results your speakers should be directional (electrostats, horns, bipoles), not like my Magicos. Or/and your need absorption. Or/and you should sit in the near-field.
This goes against my old concept in some areas, where I sought a wide off-axis response of my speakers and diffusors to create more equal room response for adding volume to the sound. I also have super-tweeters firing up / to the side to increase spaciousness. With BACCH these are counter-productive.
- BACCH interacts with the DRC filters, making them less effective. Edgar's recommendation is, that you need to extract the DRC filters with BACCH activated.
- Despite the claims to the contrary, I hear some minor changes in the frequency response/colorations with BACCH. Bass is almost unaffected, but the mids are a little recessed and the highs are more pronounced. It is not much, but clearly audible.
- the sweet spot becomes very small. The cancellation is optimized for your position. This is a limitation of the uBACCH fixed filter version. The higher versions use a webcam to track your head.
- BACCH-Intro is limited vs the full version, because the filter is not optimized for your room (can someone share experiences of using uBACCH vs full filters?)
Overall
DSP is the next level of HiFi.
DACs have become perfect, amps are ultralinear, music streamers provide endless libraries without noise. Yet, how to deal with the limitations of speakers (even though Magico, D&D, Kii and others have come a long way in smooth response without distortions and ported designs, the room modifies it)? Even worse, what to do with the issues of our rooms?
Digital room correction with Dirac/Acourate has compensated for frequency errors and timing.
BACCH eliminates another issue - the problem of cross-talk during speaker reproduction, namely that your left ear hears the right speaker (which it should not) and vice versa. It also works for headphones, removing the unpleasant issue of the "soundstage inside your head" by using head-tracking.
Floyd Toole mentioned, that of course correction is always the second best choice. The best choice would be a new sound format, where the room and frequency information is encoded and an optimized decoder regenerates the desired sound-field. While this is a great idea, it won't come anytime soon and won't work for existing recordings.
So XTC and DRC are great bets to improve your listening experience in your setup with your music library.
Save your money, don't buy expensive cables, another shiny amp, a DAC with almost the same specs, or esoteric products like holographs.
Try something new based on mathematical algorithms of sound reconstruction.
Edgar hinted, that there will be a version of BACCH coming our way with XTC and binaural room correction! That would be a killer and fix some of my above issues.
Yet, it will require the Audiophile version and break the bank again.
Introduction
My experience is, that the interaction between room and speakers has by far the greatest impact on how I enjoy music listening. Most of my upgrades of the last 10 years went into room acoustics (without completely sacrificing the looks of the room). I also finally got good results using Acourate as a digital room correction solution to compensate for the rest.
My personal findings so far match the research of Toole/Olive, namely that a flat frequency response on axis (and an even behaviour also off-axis) makes music sound "right" and for imaging you want to either absorb first reflections or diffuse, but definitely you want to diffuse late room reflections to give a sense of space. So put diffusors at the backwall and also between the speakers. A decay time RT60 of 0.25 to 0.3 seconds works very well. Below the transition frequency use bass traps and DRC. Wide dispersion speakers yield a good compromise between tonality and sound-stage. Directional and bipolar speakers like Magnepans are difficult to set up.
What BACCH-DSP does
BACCH seems to be around since 2015, developed by Prof. Edgar Choueiri of Princeton Univ. His company Theoretica Applied Physics is now more present at HiFi shows demonstrating the effect. It is a DSP that eliminates speaker cross-talk (XTC) at the listening position by a sort of a cancellation pattern (similar to noise cancellation) while avoiding frequency artifacts due to the interference. To quote from his paper: "These two series of pulses are delayed by τc with respect to each other so that after the first positive pressure pulse arrives at the left ear, then reaches the right ear with a slightly smaller amplitude, it is cancelled there by a negative pressure pulse of equal amplitude (that was emitted a time τc earlier by the right loudspeaker), which in turn is cancelled at the left ear by a positive pressure pulse, and so on. The net result is that only the first pulse is heard and only at the left ear, i.e., with no crosstalk"
It is not a new concept, but BACCH implements it very smartly. It has special algorithms for bass and mono signal treatment and lots of calculation to minimize colorations.
BACCH DSP is available in three flavors, all based on a similar/same set of algorithms
1. Expensive: BACCH4MAC intro, at $1000. You can't customize the XTC elimination filters, but can use only a generic model based on the distance between you and your speakers (uBACCH)
2. Very Expensive: BACCH4MAC audiophile $5000/$6000 with a binaural microphone + interface to measure your system's response. It also offers head tracking by webcam to maintain the benefit when you move.
3. Ultra-Expensive: BACCH Sound Processors stand-alone equipment from $20k to $50k, which you can easily plug into your signal path.
Options 1 and 2 require an Apple Mac in the signal path to run the DSP software. This adds another $1000 and honestly reduces the usability of my system: I have not yet succeeded in making the Mac headless, so I need a monitor and mouse to configure it.
You can try option 1 for 2 weeks with full refund.
Setting it up
My current setup consists of a central music server running Logitech Media Server 8.4, a raspberry Pi endpoint running SqueezeLite on AroioOS and BruteFIR DRC. This is feeding a SMSL VMV1 DAC via coax and Classé M600 monos into Magico S5 Mk2. I always wanted to avoid the hassle and - maybe irrationally - the electrical noise of a PC in the sound-chain.
It is an easy-to-use configuration. Just switch it on and select the music with the iPeng app.
When you buy the BACCH software, a service employee (in my case it was Edgar) helps you perfectly by phone to get your system running. Support of Theoretica is unbelievable! Edgar spent more than an hour with me explaining everything.
Here is what we did: set up the Apple Mac Mini M1 16GB running SqueezePlay (you can also use Roon, HQPlayer or probably almost any other software which plays music).
You need BlackHole Loop Back software to re-route the sound on the Mac from SqueezePlay to BACCH-DSP (simply create an aggregate device in the Audio/MIDI Settings with BlackHole + your DAC as sources, set the clock source to your DAC and tickmark drift compensation for BlackHole). Call this aggregate sound device "SF4BACCH". Then in the BACCH-DSP software select SFBACCH in the Auto-Configure drop-down menu. Now select a filter bin, click "ARM". enter the distances between you and your speakers and between the speakers, and click FIRE. his is more or less it, but there are some pit-falls where Edgar helped.
Another challenge was to include my DRC filters. They need to go after the XTC filtering. BACCH-DSP supports VST/AU/VST3 plugins, so you can use a Convolver VST plugin.
Selection on the Mac is huge, but some free ones don't work. Edgar recommended Reverberate (which will cost you another $99). It is not as great as BruteFir or Accourate Convolver as it doesn't change sample rate automatically for instance.
So my signal chain now looks like this:
Server (running LMS) ==> (ethernet) ==> MacMini (running SqueezePlay, BlackHole, BACCH-DSP 12.2, Reverberate Core VST3 for the DRC FIR filters) ==> (USB) ==> DAC
More complex setups with active XOs should be possible and VST plugins like Elements open further realms for post-processing.
What is the effect
Hard to describe, it creates the impression of a 3-dimensional room in front of you, musicians and sounds eminate from a precise spots in the room rather than from between the speakers, sometimes even far outside the physical dimensions of your room, or from the side/behind you or from the top.
The effect is most evident when I switch to Bypass: the musicians are pulled with a rubberband from their positions around the speakers back to a position in the speakers' plane. The 3-dimensional room collapses.
When switching it back on, the musicians slowly move to new positions in the room, sometimes almost next to you. It takes a couple of seconds for your brain to accomodate and grasp the virtual room around you. On binaural recordings of Chesky you can have people walking up close to you and whispering. It's more than a gimmick, you forget rapidly that it's there.
The magnitude varies greatly with the musical pieces. Since cross-talk is elminated your brain is re-interpreting the spatial cues which are in the recording. If there few or none, there is also no effect and no change in the presentation.
In any case, the speakers vanish. I guess this comes from the fact, that each ear gets fed with signals coming from the respective side only. So your brain has no cues from the other ear for the speaker's position.
Which music is benefitting
Naturally, the effect is biggest with binaural recordings, but many standard recordings work as well. Best is live music like a classical orchestra, electronic music sometimes is excellent, straight studio recorded pop-rock is not benefitting much in many cases. There is a thread with a list of recordings on this site, start looking there.
Limitations of BACCH
- The system complexity is higher, you need a Mac (and I don't like Macs)
- Usability of the software is astonishingly good, but the fonts are so small, that using it in a listening environment is tedious.
- The magnitude of the spatial improvement varies a lot with the music. If studio recordings with individual mics are your thing, the benefit might be less.
- For best results your speakers should be directional (electrostats, horns, bipoles), not like my Magicos. Or/and your need absorption. Or/and you should sit in the near-field.
This goes against my old concept in some areas, where I sought a wide off-axis response of my speakers and diffusors to create more equal room response for adding volume to the sound. I also have super-tweeters firing up / to the side to increase spaciousness. With BACCH these are counter-productive.
- BACCH interacts with the DRC filters, making them less effective. Edgar's recommendation is, that you need to extract the DRC filters with BACCH activated.
- Despite the claims to the contrary, I hear some minor changes in the frequency response/colorations with BACCH. Bass is almost unaffected, but the mids are a little recessed and the highs are more pronounced. It is not much, but clearly audible.
- the sweet spot becomes very small. The cancellation is optimized for your position. This is a limitation of the uBACCH fixed filter version. The higher versions use a webcam to track your head.
- BACCH-Intro is limited vs the full version, because the filter is not optimized for your room (can someone share experiences of using uBACCH vs full filters?)
Overall
DSP is the next level of HiFi.
DACs have become perfect, amps are ultralinear, music streamers provide endless libraries without noise. Yet, how to deal with the limitations of speakers (even though Magico, D&D, Kii and others have come a long way in smooth response without distortions and ported designs, the room modifies it)? Even worse, what to do with the issues of our rooms?
Digital room correction with Dirac/Acourate has compensated for frequency errors and timing.
BACCH eliminates another issue - the problem of cross-talk during speaker reproduction, namely that your left ear hears the right speaker (which it should not) and vice versa. It also works for headphones, removing the unpleasant issue of the "soundstage inside your head" by using head-tracking.
Floyd Toole mentioned, that of course correction is always the second best choice. The best choice would be a new sound format, where the room and frequency information is encoded and an optimized decoder regenerates the desired sound-field. While this is a great idea, it won't come anytime soon and won't work for existing recordings.
So XTC and DRC are great bets to improve your listening experience in your setup with your music library.
Save your money, don't buy expensive cables, another shiny amp, a DAC with almost the same specs, or esoteric products like holographs.
Try something new based on mathematical algorithms of sound reconstruction.
Edgar hinted, that there will be a version of BACCH coming our way with XTC and binaural room correction! That would be a killer and fix some of my above issues.
Yet, it will require the Audiophile version and break the bank again.
Last edited:
