BACCH4MAC Short Review / first impressions

[Gwreck]

Since I last posted in this thread, I have experimented a bit with Ambio One which is a free Ambiophonics VST plugin in JRiver. My issue is that tuning it is by ear only. I am not aware of any measurements that can be done to show that they are optimally dialled in. The effect of twiddling those knobs is actually night and day but NOT in a good way. It ranges from subtle increase in depth, width, and precision of the soundstage, to dramatic changes in these parameters but with a "phasey" and hollowed out kind of sound that frankly sounds weird. I never did get satisfactory results from Ambio One, so I am looking for a more objective approach.

I understand that Choueri worked with Ralph Glasgal when ambiophonics RACE (Recursive Ambiophonic Crosstalk Elimination) algorithm was developed. Choueri then developed BACCH. There are a number of VST's that use RACE, e.g. Ambio One (developed by Glasgal?), AmbiophonicsDSP (defunct), SoundPimp, and maybe a few others. The advantage of all those VST's are that they run on Windows and they are substantially cheaper. The disadvantage is that none of them can be set up with measurements, none have head tracking, and they all require narrow speaker angles to work properly. The lack of being able to guide your settings with measurements is what is killing the deal for me with all these non-BACCH ambio plugins.

So, I was wondering if you would be keen to do a review on BACCH from an "objectivist" perspective, with measurements and subjective listening impressions of various settings. All current reviews on BACCH are subjective, I have not found a single one that explains what settings are available and what effect it has. Not even a screenshot, apart from what you have provided.

I could do a function review with pictures. Other than BACCH and my umik-1, I use for Dirac,I don’t have any decent gear for audio measurements. (I was thinking of using REW to check for Bass response issues).

I have tried multiple versions of RACE and the only one that was even close to listenable was the mini-ambio plug-in for the MiniDSP. Even then BACCH is better is every aspect. It’s still available but was once its own special mini DSP unit. https://www.minidsp.com/products/plugins/ambiophonics-detail

 

[Keith_W]

Yes please. There are a few aspects I would be interested in which don't seem to be publicly available:

- CPU usage
- Latency
- more screenshots of the interface
- performance of head tracking in various scenarios, e.g. dim lighting, more than one listener, etc.
- whether speakers need to be repositioned or not
- whether speaker issues, such as time and phase coherency affect the effectiveness of BACCH. I know that directivity is definitely important.

I can probably think of more, but for the moment I am most curious about the general useability and what settings are available.

 

[sweetsounds]

Here is a screenshot of BACCH. It is very much, how it looks like during usage. ARM/FIRE are used for doing the measurements/filter generation, essentially using 3 log sweeps in various head positions.
You have 3 bins for different filters on the left and you can of course Bypass the effect. The middle column contains the settings, you use a virtual soundcard called BH4BACCH or SF4BACCH and which input/output channels to use. Below you can correct for mono and bass effects, I leave it on as it avoids correction artefacts for the mono content of the music.

Now, the measurement plots are hard to read for me as well, they don't match my own measurements. On the bottom there is the step response and you can see the estimated XTC.

u-BACCH goes without measurements, it is assuming an idealized speaker and calculates the filter based on your distance measurements only, it looks like on this link in the upper left side.

With the great support of @Mojo7 here is my measurement.
We don't understand the difference between left/right as my room is almost perfectly symmetric. Also the impulse response looks quite different from my own log-sweep measurements.

 

[Gwreck]

Here is a screenshot of BACCH. It is very much, how it looks like during usage. ARM/FIRE are used for doing the measurements/filter generation, essentially using 3 log sweeps in various head positions.
You have 3 bins for different filters on the left and you can of course Bypass the effect. The middle column contains the settings, you use a virtual soundcard called BH4BACCH or SF4BACCH and which input/output channels to use. Below you can correct for mono and bass effects, I leave it on as it avoids correction artefacts for the mono content of the music.

Now, the measurement plots are hard to read for me as well, they don't match my own measurements. On the bottom there is the step response and you can see the estimated XTC.

u-BACCH goes without measurements, it is assuming an idealized speaker and calculates the filter based on your distance measurements only, it looks like on this link in the upper left side.

With the great support of @Mojo7 here is my measurement.
We don't understand the difference between left/right as my room is almost perfectly symmetric. Also the impulse response looks quite different from my own log-sweep measurements.
View attachment 287702

What speakers are you using? What room treatments and furniture are in the listening area? I have found that small differences in items around me can make a large difference in the db of XTC.
I had a bookshelf sized speaker several feet behind and to the left of my head and it dropped my XTC by 10db on the right. The bottom of the chart showing impulse response is a good indicator of the amount of XTC you can achieve. As I understand it cleaner the impulse response the better.

 

[sweetsounds]

Some further information
- CPU usage on M1 Mac ~25%
- dim light is OK with a Logitech C922, Theoretica is still working on an IR camera for dark rooms
- speakers don't need to be repositioned, but recommended to toe in to point to ears and to put them closer together. At shows the speakers are roughly in a 1.2m stereo triangle pointed upwards to the listener's ears. I kept mine at 2m distance so far. Reflections should be avoided as there is no XTC for reflections and therefore the effect is less.
- directivity should be high. You want the depth information from the recordingg, not your room. My DRC also time-aligns the drivers and it adds to the effect.

Objectively measuring XTC is difficult. What we did is locating the sound by pointing to the spot where it seems to originate from in the Chesky sound demos.

Left Channel Id Test / Dr. Chesky
- No BACCH: person located at left speaker
- u-BACCH: person located 45° outside the left speaker and .5m in front of left speaker
- BACCH: person located 80°, 1.5m from left ear
- BACCH demo at shows, the person is truly whispering into your left ear.

"Counting" The Cerddorion Vocal Ensemble
- No BACCH: the voices are located in a circle at/behind the speaker plane
- BACCH demo at shows, the voices are clearly located around you, you are in the center of the circle (voices seem to come from behind as well)

 

[Keith_W]

That is very interesting. Why do you think you are unable to re-create what you hear at BACCH demos? Do you think you have a speaker issue? A speaker placement issue? A room issue? Or a user error issue?

 

[sweetsounds]

That is very interesting. Why do you think you are unable to re-create what you hear at BACCH demos? Do you think you have a speaker issue? A speaker placement issue? A room issue? Or a user error issue?

I think plenty has been written by now about the supporting and inhibiting factors for good ambisonics and immersive audio. (Experience reviews are even in the absolute sound).

My Magico speakers are directionally wide, which isn't ideal. My experiences are consistent with others using Kii, Grimm etc. If you like these, then your room needs to have low RT60 (mine is 0.25s) and no first reflection points.
XTC in bass isn't working well as its reflection can't be absorbed.

I enclose a photo of the HighEnd Munich 2023 BACCH demo on Jensen (?) Janszen speakers, where clear focus was on direct-sound-only for the listener in the sweet spot. Take note of the floor absorbers. The room was a normal meeting room and it is useful to avoid the headphones' "locked-in-feeling".

While I can't rule out a user issue (like improper seating of the ear mikes), the number of adjustments in the software for the filter generation is low, much lower than in other complex DSP tools like Acourate. (Actually I found it tricky to get BACCH to work in my configuration at all, but once it works it's OK).

Actually - and here opinions differ - I remain super-impressed with that spaciousness of 20 to 40 degrees extra width for normal music, which usually requires a bigger room or (even more) expensive speakers. I don't seek surround sound on 2 speakers (which is theoretically possible).

The technology is on the market since 2010, so it's quite mature.

 

[Gwreck]

I think plenty has been written by now about the supporting and inhibiting factors for good ambisonics and immersive audio. (Experience reviews are even in the absolute sound).

My Magico speakers are directionally wide, which isn't ideal. My experiences are consistent with others using Kii, Grimm etc. If you like these, then your room needs to have low RT60 (mine is 0.25s) and no first reflection points.
XTC in bass isn't working well as its reflection can't be absorbed.

I enclose a photo of the HighEnd Munich 2023 BACCH demo on Jensen (?) speakers, where clear focus was on direct-sound-only for the listener in the sweet spot. Take note of the floor absorbers. The room was a normal meeting room and it is useful to avoid the headphones' "locked-in-feeling".

While I can't rule out a user issue (like improper seating of the ear mikes), the number of adjustments in the software for the filter generation is low, much lower than in other complex DSP tools like Acourate. (Actually I found it tricky to get BACCH to work in my configuration at all, but once it works it's OK).

Actually - and here opinions differ - I remain super-impressed with that spaciousness of 20 to 40 degrees extra width for normal music, which usually requires a bigger room or (even more) expensive speakers. I don't seek surround sound on 2 speakers (which is theoretically possible).

The technology is on the market since 2010, so it's quite mature.

I am using Sander’s ESL’s in a large basement family room. The only acoustic treatments are 4 inch thick absorption panels on the wall behind each of the speakers. My XTC is 20+ db which apparently is the amount needed for maximum effect. I have also successfully used an intel IR camera for BACCH and it works well. Seeing the red dots all over on video is bizarre though. The biggest issue I have with BACCH in general is when I don’t have good head tracking which creates a jumping of positions. The sound effect is almost that of tape flutter. More light and less distance fix these issues.

 

[Dialectic]

I am using Sander’s ESL’s in a large basement family room. The only acoustic treatments are 4 inch thick absorption panels on the wall behind each of the speakers. My XTC is 20+ db which apparently is the amount needed for maximum effect. I have also successfully used an intel IR camera for BACCH and it works well. Seeing the red dots all over on video is bizarre though. The biggest issue I have with BACCH in general is when I don’t have good head tracking which creates a jumping of positions. The sound effect is almost that of tape flutter. More light and less distance fix these issues.

I consistently get 10-14 dB of XTC with the Dutch & Dutch 8Cs. Plenty for breathtaking stereo imaging and a sense of envelopment, not quite enough for crazy, behind-the-head effects.

 

[sweetsounds]

I am using Sander’s ESL’s in a large basement family room. The only acoustic treatments are 4 inch thick absorption panels on the wall behind each of the speakers. My XTC is 20+ db

Could you give us some more information on the setup, since you achieved the best XTC:
- speaker model? 10c/10e?
- Distance between loudspeakers
- Distance to listening position
- Distance loudspeakers to rear and side walls
- Toe in / speaker orientation

Are you using a DSP to correct for the speaker frequency response?
I don't know about the ESL series, but I hope that they have a better frequency response than their predecessors and they probably remained amazing "beamers":

Stereophile: InnerSound Eros Mk.III electrostatic loudspeaker

 

[Gwreck]

Could you give us some more information on the setup, since you achieved the best XTC:
- speaker model? 10c/10e?
- Distance between loudspeakers
- Distance to listening position
- Distance loudspeakers to rear and side walls
- Toe in / speaker orientation

Are you using a DSP to correct for the speaker frequency response?
I don't know about the ESL series, but I hope that they have a better frequency response than their predecessors and they probably remained amazing "beamers":

Stereophile: InnerSound Eros Mk.III electrostatic loudspeaker

-10e
-the speakers are 10ft from my listening position on a 40 degree arc. I laser measure each side of each speaker to ensure the distance is the same to my position. I use the light test also to ensure the speakers panel is in the same orientation to me. I don’t know the distance between the speakers.
-The speakers panel is about 3 Ft off the back wall.
- I was using the DBX venue 360 with the sanders which also corrected for frequency response. I am now using the RME UCX II for BACCH and the preamp/crossover for the Sanders. The UCX has eq features but I am not using it. BACCH has some graphic eq features but I also am not using it currently.

Here is picture of the layout. It’s a bit of a mess. I currently have a 98inch TV on the wall and my equipment on a stand below. Those are 2x4 4inch thick acoustamac panels. On the lower half of the wall I have 2x2x2in similar panels behind the speakers(I don’t think they do much).

 

[Gwreck]

I consistently get 10-14 dB of XTC with the Dutch & Dutch 8Cs. Plenty for breathtaking stereo imaging and a sense of envelopment, not quite enough for crazy, behind-the-head effects.

I someday would like to hear a pair of those speakers. I have only heard good things.

 

[sweetsounds]

-10e
-the speakers are 10ft from my listening position on a 40 degree arc. I laser measure each side of each speaker to ensure the distance is the same to my position. I use the light test also to ensure the speakers panel is in the same orientation to me. I don’t know the distance between the speakers.

That helps, interestingly, that absorption is rather minor and speakers are close ot the wall, also you have a high floor carpet, which might help.
Could you post a screenshot of BACCH? Would like to compare the response/step with my values.

 

[Gwreck]

That helps, interestingly, that absorption is rather minor and speakers are close ot the wall, also you have a high floor carpet, which might help.
Could you post a screenshot of BACCH? Would like to compare the response/step with my values.

The very directional nature of the Sanders makes them more immune to room interactions. Dr Choueri has tested many speakers and the sanders are the most directional. He has made a custom rear wave absorber. I just use wall treatments. Even with my numbers my seating position is not ideal as I have a high backed leather sectional. Fortunately my spouse’s obsession with throw pillows and furs on the couch is helpful. My room is approximately 35ft in each direction and I have 9 ft ceilings in the basement room. My seating position is midway on one wall and 13-14 Ft from the other.

 

[STC]

Left Channel Id Test / Dr. Chesky
- No BACCH: person located at left speaker
- u-BACCH: person located 45° outside the left speaker and .5m in front of left speaker
- BACCH: person located 80°, 1.5m from left ear
- BACCH demo at shows, the person is truly whispering into your left ear.

Thank you for the description but I have few more questions.

My reference to the degrees are:-
0 degree - straight ahead or midpoint of the two speakers.

90 degrees - at ether 9 or 3 o clock to us.

My understanding is the voice should emerge about 30 degrees from the left which is the typical stereo equilateral triangle. In 2012, I don’t think anyone would have made a test CD tracks meant for BACCH in 2012 or any kind of XTC type playback. So my question is should you be hearing the voice coming from 90 degrees (9 o’clock) or about 30 degrees ( 11 o’clock) from the mid point of the two speakers?

I don’t think Chesky’s would have made the test CD to produce the sound to come from 9 o clock as it would have been impossible to be produced by the speakers. So I am wondering if the BACCH producing the voice at 80 degrees ( or about 9 to 10 o’clock is probably due to incorrect setting.

Using the settings derived from my own reference recordings for my crude RACE based XTC (without head tracking, I observed the followings:-

At the beginning,

MyXTC - it is around 30 to 40 degrees. The recording contains a lot of reverberation so I was expecting a fixed position unlike my reference recording.

Then when he starts to walk it slightly moved about 45 to 50 ( around 10 o’clock ). My guess is he started to walk straight ahead or the change in reverbs destroyed the accurate localization.

After that, it moved back to 30 degrees ( 11 o’clock) and continued towards me. The whisper was around the arm’s length but some listeners perceive them to be at the ears. I can’t explain this because even my own reference track with buzzing bees I cannot perceive them at my ears but there are some listeners swear it was going inside their ears. I guess localization also involves prior knowledge and a lot of imagination and good hearing.

The BACCH at Demo didn’t say where the sound emerged from but the other BACCH you are saying that the sound emerged almost to your immediate left makes me wonder which one should be correct?

 

[sweetsounds]

Thank you for the description but I have few more questions.

My reference to the degrees are:-
0 degree - straight ahead or midpoint of the two speakers.

90 degrees - at ether 9 or 3 o clock to us.

My understanding is the voice should emerge about 30 degrees from the left which is the typical stereo equilateral triangle.

Actually, no. With uBACH it would be at 10 o'clock in your coordinates, 0.5m in front of the speaker plane and left of the left speaker.
With BACCH audiophile it is at "9:30" 1.5m away
With full XTC as during the demo should be at 9 o'clock about 10cm from your left ear.
The voice "whispers into your ear", which is quite shocking.

The voice ensemble recording counting numbers is intended to be behind you as well. In the demo I heard them at 5 o'clock about 1m behind my head.

In 2012, I don’t think anyone would have made a test CD tracks meant for BACCH in 2012 or any kind of XTC type playback.

Chesky made binaural recordings to be listened to over headphones and the concept is more than 100 years old. BACCH brings it to speakers.
If XTC is fully working, the sounds should be coming in full 3D and not be limited in between or behind the speakers.
Choueiri and Chesky worked together on some starting 2010.

Notice: it is much more difficult to position objects above your head, because your ears and the speakers are in the same plane.



According to your description, MyXTC doesn't achieve the right projection in your setup.
In mine the voice is always outside the speaker plane. Even at the beginning it is far behind my rear wall and 2m left of the left speaker.

The BACCH at Demo didn’t say where the sound emerged from but the other BACCH you are saying that the sound emerged almost to your immediate left makes me wonder which one should be correct?

The way it was recorded was with a binaural microphone setup, most likely with silicone pinnae and Chesky went up to the left microphone to whisper.
You should hear exactly this. Every deviation from it comes from incomplete XTC.

 

[STC]

Actually, no. With uBACH it would be at 10 o'clock in your coordinates, 0.5m in front of the speaker plane and left of the left speaker.

Firstly, the Chesky binaural recordings are not true binaural recordings. They use filter that was developed by Choueiri for loudspeakers in playback. Later Chesky binaural recordings are called binaural + to reflect they are also playable on loudspeakers.

In my opinion, the sound should merge from 30 feet away from the listener. If your speakers are placed 30 feet away then one would say it is emerging right from the speaker. Although, IMO human judge distance by the reverbs and loss of HF and without an actual reference you are just imagining the location based on the info provided.

There is no rule that sound must always come far away from the speaker. The sound emerges from the location based on the ILD and ITD ( also reverbs for distance) that your brain could decipher from the recording.

In my case the speakers are placed touching one another in what’s known as Ambiodipole configuration of 19 degrees angle. Therefore all reference are outside the left speaker.

With BACCH audiophile it is at "9:30" 1.5m away
With full XTC as during the demo should be at 9 o'clock about 10cm from your left ear.

Again, this is impossible and they would not have done so because BACCH was not available then and it serves no purpose even for binaural with headphones recordings. I have seen multiple recording session by Chesky using the dummy head and the always confine the stage to somewhat about 60 or 90 degree.

The voice ensemble recording counting numbers is intended to be behind you as well. In the demo I heard them at 5 o'clock about 1m behind my head.

My reference is only the left channel ID track.

Chesky made binaural recordings to be listened to over headphones and the concept is more than 100 years old. BACCH brings it to speakers.

Chesky’s binaural recording are made for both using a filter to do the correction. BACCH or XTC purpose is for something else and not confined to binaural recordings.

If XTC is fully working, the sounds should be coming in full 3D and not be limited in between or behind the speakers.
Choueiri and Chesky worked together on some starting 2010.

Choueiri was one of the engineer hired by Ralph Glasgal to develope his Ambiophonics. I forgot the exact timeline but I believe it was 2008. Choueiri went on to develope a better XTC. I think the initial stage when it was under Princeton Uni it was called Pure Audio and Sony did give a grant for the research.

I don’t think BACCH ever made a claim that it could produce a full 3D. Putting sound behind the head is not easy ( Ralph said if one could that they are audio Einstein ) .I would like to see the papers claiming that could produce sound behind the head. Occasionally, you can get sound behind the head even with stereo but that is something else. Even in those days when I was playing stereo, one Madonna track have an effect the sound moving from front to back . I have listen to the track again to see how the effect is produced in my Ambiophonics system. Been years since I heard them. Hehehe

According to your description, MyXTC doesn't achieve the right projection in your setup.
In mine the voice is always outside the speaker plane. Even at the beginning it is far behind my rear wall and 2m left of the left speaker.

You could be right! But you were not there so you have no idea how they were positioned at the time of the recording. I have binaural mics and Made my own recording so that I know exactly where the sound should emerge. It is not that I couldn’t make it to emerge at 9 o’clock but that would be wrong. So if my own recording could produce the location exactly like how I recorded them then why should Chesky ID be different? That’s an interesting question.

The way it was recorded was with a binaural microphone setup, most likely with silicone pinnae and Chesky went up to the left microphone to whisper.
You should hear exactly this. Every deviation from it comes from incomplete XTC.

I am very well familiar with their recording and the dummy head they used. The question is who is correct. Unless you have binaural mics and make your own recording then you have no reference. You just take it to be correct based on what you are told or heard which may not be correct in the first place.

 

[Gwreck]

Actually, no. With uBACH it would be at 10 o'clock in your coordinates, 0.5m in front of the speaker plane and left of the left speaker.
With BACCH audiophile it is at "9:30" 1.5m away
With full XTC as during the demo should be at 9 o'clock about 10cm from your left ear.
The voice "whispers into your ear", which is quite shocking.

The voice ensemble recording counting numbers is intended to be behind you as well. In the demo I heard them at 5 o'clock about 1m behind my head.






Chesky made binaural recordings to be listened to over headphones and the concept is more than 100 years old. BACCH brings it to speakers.
If XTC is fully working, the sounds should be coming in full 3D and not be limited in between or behind the speakers.
Choueiri and Chesky worked together on some starting 2010.

Notice: it is much more difficult to position objects above your head, because your ears and the speakers are in the same plane.



According to your description, MyXTC doesn't achieve the right projection in your setup.
In mine the voice is always outside the speaker plane. Even at the beginning it is far behind my rear wall and 2m left of the left speaker.




The way it was recorded was with a binaural microphone setup, most likely with silicone pinnae and Chesky went up to the left microphone to whisper.
You should hear exactly this. Every deviation from it comes from incomplete XTC.

The Chesky recordings are made with a generic head. Positional differences may come down to differences between the generic head and the listeners head not just incomplete XTC.

 

[STC]

The Chesky recordings are made with a generic head. Positional differences may come down to differences between the generic head and the listeners head not just incomplete XTC.

True but the changes would not be too big a difference.

 

[sweetsounds]

Firstly, the Chesky binaural recordings are not true binaural recordings. They use filter that was developed by Choueiri

In my opinion, the sound should merge from 30 feet away from the listener. If your speakers are placed 30 feet away then one would say it is emerging right from the speaker.

You are right that we don't exactly know, how the recording situation was. I assume that the intend was close to the BACCH demo with sound at 09:00.
The reason I referenced to the speaker position, because w/o XTC the sound is located at the speaker for hard-panned sounds.

I don’t think BACCH ever made a claim that it could produce a full 3D. Putting sound behind the head is not easy ( Ralph said if one could that they are audio Einstein )

As you described some Qsound (Roger Waters) on some speakers (Cube Nenophar) did that. Not sure if it was intended or more 10 o'clock as I hear it in many systems.

Anybody's brain is capable to clearly hear sounds from behind. I guess it is very hard to artificially position them in studio mixing.
The only thing I can state is that in the vocal recording some voices came from behind.

There is a BACCH 3DM module which makes this claim, but I have no experiences.

So if my own recording could produce the location exactly like how I recorded them then why should Chesky ID be different? That’s an interesting question.

Correctness is a difficult term in any reproduction (to me it's mostly illusive, "good and real" sounding are more important to me, Hollywood movies even rely on artificial sounds, but some sound more real then reality orhers are gimmicky).

What I like about BACCH is the opening effect of the soundstage, a widening, deepening, and sharpening of where I can locate instruments.
I don't listen to binaural recordings and find many of them boring.

The original question was, if there are ways to measure the effect. I offered the left ID recording with positioning the sound as a reference.