BACCH, Ambiophonics, etc

[AJ Soundfield]

6 hours ago, Ralph Glasgal said:

Here is a better correction.  One can measure one speaker and then tell the reader he is on his own in judging from the data how that speaker pair is going to work in a system.  But speakers have more than just frequency response, etc.  They may have large variations in dispersion, symmetry, etc. which will affect their frequency response when they are used in pairs and the response is measured at about 4 inches off the center line or even at a few such points.  Maybe such a speaker pair would have flatter response at 57 or 64 degrees.  So that is one reason why I try to make audiophiles aware of combing and crosstalk.     

Well, here is just one data point, stereo 60 then ambio with PEQ

 

 

While in my totally uncontrolled listening, the ambio sounds a bit "clearer", it is far from night and day, much more subtle. At some point if I find the time, I will use identical pairs of speakers at 60 and 20 (same radial distance), level match and quick switch. If I do it with the audio club I may even throw in blinding.

[semente]

12 hours ago, fas42 said:

 

Yes, the response won't be flat - but the key to achieving subjectively satisfying replay is 'tricking' the ear/brain mechanism into doing the DSP for one. This is something I've spent many years exploring, and a key aspect is that subjectively there is an on/off switch inside your head that is triggered, depending upon the quality of the sound - this most likely won't be the case for everyone, but every indication is that it is largely universal.

 

Alice's mushrooms are legal in Holland and they're said to make wonders in that department. 

[Ralph Glasgal]

14 hours ago, AJ Soundfield said:

Well, here is just one data point, stereo 60 then ambio with PEQ

 

 

While in my totally uncontrolled listening, the ambio sounds a bit "clearer", it is far from night and day, much more subtle. At some point if I find the time, I will use identical pairs of speakers at 60 and 20 (same radial distance), level match and quick switch. If I do it with the audio club I may even throw in blinding.

I confess I really don't know what I am looking at, but the lower curve sure looks flatter to me.  You do need to compare the response with speakers at 60 degrees using stereo and speakers at say 15 degrees using RACE with the miniambio controls correctly set for the 15 degree angle.  Likely -2.7 dB attenuation and maybe 70 microseconds but I am just guessing.  If you can really do this comparison you will be the first one to have ever published such a fair  test.

[AJ Soundfield]

The top graph is 7' stereo 60, L, R (bottom 2 curves), then summed LR top curve

The bottom graph is 7' ambio 20 summed LR with PEQ applied, 2 mic positions, one about 8" back from other. All are 1/12th octave resolution.

If I were to apply psychoacoustic smoothing, they would all be visually smoother, as the ears resolution is lower.

There is no drastic difference between the stereo and ambio. Very subtle

[Ralph Glasgal]

50 minutes ago, AJ Soundfield said:

The top graph is 7' stereo 60, L, R (bottom 2 curves), then summed LR top curve

The bottom graph is 7' ambio 20 summed LR with PEQ applied, 2 mic positions, one about 8" back from other. All are 1/12th octave resolution.

If I were to apply psychoacoustic smoothing, they would all be visually smoother, as the ears resolution is lower.

There is no drastic difference between the stereo and ambio. Very subtle

I did remember a similar test conducted at NYU.  It includes SRS in the comparison but has a variety of test conditions.  More such tests need to be done with a variety of speakers, angles, etc.  Maybe the Europeans someday.

Fig. 7: The measured response. red: digital file; blue: reference stereo; green: SRS iWOW orange: Ambidio  You can see that the Ambio (Ambidio) orange curves are flatter.  The principle is that the better the crosstalk cancellation the flatter the response becomes.  So you could look at the response on a scope and adjust the MiniDSP box for the flattest response at an ear at the listening position.,

[semente]

5 hours ago, Ralph Glasgal said:

I did remember a similar test conducted at NYU.  It includes SRS in the comparison but has a variety of test conditions.  More such tests need to be done with a variety of speakers, angles, etc.  Maybe the Europeans someday.

Fig. 7: The measured response. red: digital file; blue: reference stereo; green: SRS iWOW orange: Ambidio  You can see that the Ambio (Ambidio) orange curves are flatter.  The principle is that the better the crosstalk cancellation the flatter the response becomes.  So you could look at the response on a scope and adjust the MiniDSP box for the flattest response at an ear at the listening position.,

 

I would like to listen someday, just so I can comment from experience.

My understanding tells me that a physical partition between speakers would not also provide better crosstalk cancellation but would also do away with weird, dizzying effects; perhaps I am wrong.

Let me know when you next demo in London.

[AJ Soundfield]

9 hours ago, Ralph Glasgal said:

I did remember a similar test conducted at NYU.  It includes SRS in the comparison but has a variety of test conditions.  More such tests need to be done with a variety of speakers, angles, etc.  Maybe the Europeans someday.

Fig. 7: The measured response. red: digital file; blue: reference stereo; green: SRS iWOW orange: Ambidio  You can see that the Ambio (Ambidio) orange curves are flatter.  The principle is that the better the crosstalk cancellation the flatter the response becomes.  So you could look at the response on a scope and adjust the MiniDSP box for the flattest response at an ear at the listening position.,

Ralph, those are laptop speaker measurements! That's why they band limited to 100-10k. It's all in the paper

http://www.ambiophonics.org/papers/AES136IRIS.pdf

[Ralph Glasgal]

10 hours ago, semente said:

 

I would like to listen someday, just so I can comment from experience.

My understanding tells me that a physical partition between speakers would not also provide better crosstalk cancellation but would also do away with weird, dizzying effects; perhaps I am wrong.

Let me know when you next demo in London.

Weird, dizzying effects?  Is that an urban myth like the one about stereo reproduction being flat? 

[Ralph Glasgal]

7 hours ago, AJ Soundfield said:

Ralph, those are laptop speaker measurements! That's why they band limited to 100-10k. It's all in the paper

http://www.ambiophonics.org/papers/AES136IRIS.pdf

I always thought they were done with other speakers but rereading the thesis I think you are correct.  They did not have low bass, no subwoofer, etc. so when they asked me to review the paper, I told them not to include that band in the paper as it could not really have been audible.  At the high end there is no crosstalk or anything that makes sense in comparing RACE to stereo or SRS so again since the level of the test signal coming out at HF was dubios they agreed to limit their conclusions to the band where things really matter.   Also RACE only does anything between 100 and maybe 9kHz or so and that is what the laptop speakers could clearly cope with. 

 

I can assure you that the results are the same whether the speakers are of audiophile quality or not.  It reminds me of the tests poopooed in desperation by announcing that the switches used to do the double blind testing invalidated the results.  I think the miracle of this thesis is that a human listening panel could hear things like realism, depth, etc. from a laptop.  If it had just been done with big box speakers, nobody would care since that is easy.

[AJ Soundfield]

30 minutes ago, Ralph Glasgal said:

I can assure you that the results are the same whether the speakers are of audiophile quality or not. 

No you can't, because the test isn't remotely connected a typical stereo 60 triangle.

This is the equivalent of testing stereo at something like 20 degrees which is specific to ambio.

I have no doubt ambio works great for built in laptop speakers 12" apart.

 

Quote

It reminds me of the tests poopooed in desperation by announcing that the switches used to do the double blind testing invalidated the results.

That's silly. This is a completely invalid test for stereo 60 vs ambio 20 for typical loudspeakers

[Ralph Glasgal]

1 hour ago, AJ Soundfield said:

No you can't, because the test isn't remotely connected a typical stereo 60 triangle.

This is the equivalent of testing stereo at something like 20 degrees which is specific to ambio.

I have no doubt ambio works great for built in laptop speakers 12" apart.

 

That's silly. This is a completely invalid test for stereo 60 vs ambio 20 for typical loudspeakers

A fixed dummy head was used to measure all the frequency responses.  So as far as I know all the measurements were made at roughly 60 degrees since this is a special version of RACE for that angle.  More importantly in the other tests listeners heads were also fixed at the same angle for all the tests and if anything stereo was at an advantage.  I wish, as I said, I had listening panels to spare but I do have lots of measurements with a variety of speakers most of which you have seen and as time goes on and Ambio becomes more popular there will be more reviews.   I think I posted two review links here.

 

There are a whole bunch of kudos from users at www.ambiophonics.org/Kudos.html  including Clark Johnsen, but that is not the AES way to prove a technology since they used ears and not objective methods.

 

Why don't you do the test at 60 stereo and 20 RACE degrees with a simple mic at 4 inches off the center line and tell us what you find.   

[AJ Soundfield]

12 minutes ago, Ralph Glasgal said:

A fixed dummy head was used to measure all the frequency responses.  So as far as I know all the measurements were made at roughly 60 degrees since this is a special version of RACE for that angle. 

No. That would require one's face on the computer to form an equilateral 60 triangle with laptop speakers 12" or so apart. I'm having doubts you have read the papers you linked!

These papers simply do not support anything other than what they tested laptop speakers using ambio vs SRS.

Attempting to extrapolate that to stereo 60 loudspeakers in a reflective living room is pure folly.

 

Quote

Why don't you do the test at 60 stereo and 20 RACE degrees with a simple mic at 4 inches off the center line and tell us what you find.   

What do you think I'm measuring??

[Ralph Glasgal]

3 minutes ago, AJ Soundfield said:

No. That would require one's face on the computer to form an equilateral 60 triangle with laptop speakers 12" or so apart. I'm having doubts you have read the papers you linked!

These papers simply do not support anything other than what they tested laptop speakers using ambio vs SRS.

Attempting to extrapolate that to stereo 60 loudspeakers in a reflective living room is pure folly.

 

What do you think I'm measuring??

The angle may have been less than 60 degrees but that favors stereo frequency response since the combing starts at a higher frequency where it is less troublesome to the pinna.  The human listeners all heard a stereo stage so the angle could not be all that prejudicial and SRS worked.

 

I should point out that nowhere is there any definitive paper proving that 60 degrees is optimum.  Not mathematically or psychoacoustically  That is why you can have speakers all over the place in cars, computers, clock radios, etc.  Same is true for Ambio, I cannot prove that crosstalk cancellation is perfect if you do it at 15.39 degrees.  Again the NYU test listeners all claimed to hear a normal stereo stage.  Again, if you don't care about stage width, stereo is flatter if you move the speakers closer together.  

 

But the main point, that can be proven mathematically, is that two speakers at say 40 t0 60 degrees radiating  any energy at frequencies in common will produce interference patterns everywhere except on the exact line between them.  The interference or combing pattern for 60 degrees was published in AES Preprint 2420A and 2420B Nov. 1986 entitled "The effects of interaural crosstalk on stereo reproduction and minimizing interaural crosstalk in nearfield monitoring by the use of a physical barrier by Don B. Keele Jr. and Tim Bock.  I published at the same time unaware of their previous paper.

 

You could confirm their results for the modern non-AES world and all audiophiles using your speakers with high rez test signals and a great mic.  Maybe the combing and bass boost would be less at 75 degrees or at 50 with some kind of shaping dispersion.

 

The room was normal for NYU.  Are you claiming that since they listened in the nearfield and room reflections were thus not as loud that that makes the test biased?  There are countless AES papers on the effects of room reflections on reproduction.  But this paper mostly dealt with detailed frontal localization parameters.  I hope they will repeat it with surround parameters included, like envelopment.

 

Great discussion though.  Perfection is hard for me now.

 

[fas42]

3 hours ago, Ralph Glasgal said:

I can assure you that the results are the same whether the speakers are of audiophile quality or not.  It reminds me of the tests poopooed in desperation by announcing that the switches used to do the double blind testing invalidated the results.  I think the miracle of this thesis is that a human listening panel could hear things like realism, depth, etc. from a laptop.  If it had just been done with big box speakers, nobody would care since that is easy.

 

Amusingly, I've done this successfully with my current laptop. The previous unit, a top end Dell, had faulty sound and it was impossible to do anything to make it pleasant in any fashion; the old HP I'm now using has a big sticker on it, Dolby HOME THEATER - IOW, they took a bit more care to get the sound components working reasonably.

 

And it shows. Realism and depth is quite straightforward for this unit, just on standard stereo tracks - I've worked out the optimum software configuration and hardware settings to get the most out of it. If I want the full immersive effect I lower my head so my chin is resting on top in the middle of the keyboard, the speakers are now very close to being in a headphone configuration with respect to my ears - and, it works ...

[Keith_W]

Sorry for the gravedig. I am keen to try Ambiophonics on my PC running JRiver, but I have hit a few snags. 

 

First, the paid version - AmbiophonicDSP VST, available here from electro-music, seems to have some issues with buyer satisfaction. Namely, their forum has multiple people complaining that they paid for the plugin and did not receive it. Apparently, the resolution process is to complain on the forum, then an admin will ring the guy and remind him to fulfil his obligations. Also, I attempted to purchase the plugin (before I read their forums), and their shopping cart does not work. So, even if this was possible to purchase, there is a chance your purchase may not be fulfilled. 

 

Next, the free version - Ambio, is almost impossible to find for download. I presume the author of the plugin is Weldroid - if you go to that page, you will find two download links. Both of them don't work. Nevertheless, after a bit of googling I was able to download Ambio from here. 

 

That was when I encountered my next issue. JRiver (I am using the current version, MC30) doesn't recognize it at all. Reading the Weldroid blog page, it appears that the Ambio plugin is 32 bit and may not work well with 64 bit VST hosts. There are VST wrappers for Foobar2000 and WinAmp, but not for JRiver. The links given for the Foobar and WinAmp wrappers do not work either.

 

One way around it would be route output through software like Audiomulch, but that costs USD$189. I downloaded the trial version of Audiomulch, which was able to load the Ambio VST plugin, but unable to send live audio to it from JRiver. 

 

So after all that, the end result is: unable to use Ambio. Is anybody still using it in 2023? How did you get it to work? 

[STC]

10 hours ago, Keith_W said:

Sorry for the gravedig. I am keen to try Ambiophonics on my PC running JRiver, but I have hit a few snags. 

 

First, the paid version - AmbiophonicDSP VST, available here from electro-music, seems to have some issues with buyer satisfaction. Namely, their forum has multiple people complaining that they paid for the plugin and did not receive it. Apparently, the resolution process is to complain on the forum, then an admin will ring the guy and remind him to fulfil his obligations. Also, I attempted to purchase the plugin (before I read their forums), and their shopping cart does not work. So, even if this was possible to purchase, there is a chance your purchase may not be fulfilled. 

 

Next, the free version - Ambio, is almost impossible to find for download. I presume the author of the plugin is Weldroid - if you go to that page, you will find two download links. Both of them don't work. Nevertheless, after a bit of googling I was able to download Ambio from here. 

 

That was when I encountered my next issue. JRiver (I am using the current version, MC30) doesn't recognize it at all. Reading the Weldroid blog page, it appears that the Ambio plugin is 32 bit and may not work well with 64 bit VST hosts. There are VST wrappers for Foobar2000 and WinAmp, but not for JRiver. The links given for the Foobar and WinAmp wrappers do not work either.

 

One way around it would be route output through software like Audiomulch, but that costs USD$189. I downloaded the trial version of Audiomulch, which was able to load the Ambio VST plugin, but unable to send live audio to it from JRiver. 

 

So after all that, the end result is: unable to use Ambio. Is anybody still using it in 2023? How did you get it to work? 

Use JBridge to convert 32 bit VST to 64bit. I used AmbiophonicsDSP with JRiver but now I am using Reaper and doing the XTC there. 
 

Unfortunately, AmbiophonicsDSP is no longer being sold. Horowitz and Miller were the designers of the plug-in. Miller got his own 3D system under Filmmaker so you may want to ask him about the plugin. 
 

The other way to run your 32 bit plug-in is to download 32 bit JRiver. In fact, the 32 bit is more stable and more than enough for our needs. 
 

The best implementation of Ambio is via Neutron Media Player with highest possible resolution. This is a high end media player and the RACE plugin is under the surround section. 
 

You don’t need Audiomulch although it is easier to use for Ambio and domestic concert hall. You can achieve the same via DAW such as Reaper. 
 

Soundpimp also sells the own plugin. They also have their own player with Ambio implemented. 
 

Of course, now you have other ways to get 3D sound of Ambio by using Apple spatial but currently the camera head tracking is confined to their device. BAACH camera head tracking probably is the next best option. 
 

Ambio is out there but under different name and flavor. Try searching for 3D audio or spatial sound. Polk SDA is also Ambio but without the flexibility to adjust the parameter to fit your HRTF. 

[Keith_W]

Thank you for your response. 

 

I downloaded JBridge and got it to convert Ambio from 32 bit to 64 bit. I went into JRiver to install it, but it promptly crashed. Every attempt I made to get it to work crashed JRiver. I was unable to access JRiver's DSP studio to remove the plugin, so I uninstalled JBridge. That allowed me to get back into JRiver and remove the now orphaned Ambio plugin. Since it's so badly behaved I did not want to try to purchase it. 

 

So I looked to see if there were any other 32 bit to 64 bit VST converters. I found this page, but some of them were either unavailable or required purchase without trial. 

I emailed Soundpimp yesterday requesting a trial of their software, but I have not heard back from them so far. We'll see what happens. 

 

Neutron Media Player appears to be Android only. I am using Windows 11. 

 

I will try your other suggestions - 32 bit JRiver, and do some googling. 

[Keith_W]

OK! So the 32 bit version of JRiver worked. The Ambio one VST plugin sounds really strange to my ears though. The tonality is noticeably different, it seems to depress the midrange and make my system sound a bit like a karaoke machine. Although the soundstage is wider, soloists seem to be stretched out across the space. 

 

My speakers are set up at for standard equilateral listening. I have NOT yet repositioned them to the 20deg angle recommended by Ambiophonics. To do so is a major undertaking because they are extremely heavy monsters. I will try it later and report my results. 

[STC]

@Keith_W NeutronMP is Window compatible. I used it with my window and iPhone. See here https://neutroncode.com/technology

 

What you experienced with Ambio DSP was due to wrong settings. Even the AmbiophonicsDSP of Miller’s guide was not clear and their default setting was wrong. It should have started with the space setting at Max. It causes the sound to be thin and tonally weird. My suggestion is get the better ones with Zentrum/ space function. Not all stereo sound is stereo, some parts in it are mono and it should not be cancelled.  Once set correctly, it should sound similar with stereo but with the 3D effect. 
 

 

[STC]

18 hours ago, Keith_W said:

OK! So the 32 bit version of JRiver worked. The Ambio one VST plugin sounds really strange to my ears though. The tonality is noticeably different, it seems to depress the midrange and make my system sound a bit like a karaoke machine. Although the soundstage is wider, soloists seem to be stretched out across the space. 

 

My speakers are set up at for standard equilateral listening. I have NOT yet repositioned them to the 20deg angle recommended by Ambiophonics. To do so is a major undertaking because they are extremely heavy monsters. I will try it later and report my results. 

It shouldn’t. Stereo and Ambiophonics sample as heard in my system. Recorded with DPA binaural mics. Use headphones.