SQ Shadow Vector Soundfield Mapping

[Lynn Olson]

Reflecting on the last paragraph of the previous post, it is a little strange that products as different as a dynamic decoder, loudspeakers, and vacuum-tube amplifiers would all have common sonic traits. They don’t share any technology, after all, and the operating principles are completely different.

I asked my partner, Karna, last night what they all had in common, since it’s actually very difficult for me to describe what my designs sound like. She’s not an audiophile, so she doesn’t use audiophile lingo, so I was curious what her impressions were (and she’s had 30+ years of listening to my prototypes). Her answer was interesting and unexpected: she said “multidimensional”, and she meant it in the broadest sense of the word.

The more I thought about it, the more I had to agree. It must emerge from an unconscious design esthetic, because, again, all three product classes use very different technology, and work in very different ways. But everything I’ve done, from Shadow Vector, to the Audionics range of loudspeakers, to the Ariel, and the Amity and Karna all sound like I’ve designed them.

For that matter, my collaboration with Thom Mackris of Galibier Designs on his NiWatt 300B SET amplifier resulted in a strong flavor of the Karna amplifier showing up in it, despite radically different topologies. (The NiWatt is a fairly conventional capacitor-coupled single-ended amplifier, while the Karna is an unusual all-transformer-coupled fully balanced amplifier.) Yet on direct A/B comparison, which Thom and I have done, they are almost indistinguishable from each other, while being quite different compared to other commercial amplifiers. (I hope that Thom goes ahead and manufactures the NiWatt amplifier; it’s a very good example of its type.)

So there’s a subtle design esthetic that’s there, stretching from 1972 to the present. Malcolm will hear it, and it will also be there in the all-software Shadow Vector decoder. It’s kind of at right angles to the mainstream high-end audio esthetic, which has gone off in a very different direction.

The mainstream has been pursuing something they call “accuracy” and “detail” since the Eighties, while I’ve been interested in depth, spaciousness, and a natural you-are-there feeling, with the singer right in the room with you. The sound should be vivid with a strong sense of physical presence. (Hint: if you start feeling a little trippy after 10 to 15 minutes of listening, that’s perfectly normal. No, I don’t know how that sneaks in there, but it does.)

Those are the esthetic goals.

 

[malcolmlear]

Hi Lynn,
Indeed, even though the new hardware based decoder criteria has been mathematical accuracy from input to output, it most certainly possesses that wonderful depth and spaciousness you mentioned which I'm sure is all down to the concept of the 4 locked rotating vectors. I did consider the idea of a SQ fallback to a Scheiber tetrahedron, however the equations to my eye quickly became 'ugly' and forced, so I reverted to your original idea. I've always found effective solutions as with nature tend to be simple.

 

[malcolmlear]

For anyone out there developing quad stuff here are a couple of test file creation programs compiled for DOS. The first creates a tone that can rotate around any of the 3 Scheiber sphere axes (good for an Olson phase spin test). The second is a fairly comprehensive test tone generator for most quadraphonic encode methods.

 

[Lynn Olson]

Hi Malcolm, I’m so glad your subjective experience matches my memories of the SV hardware prototype. Although a static SQ decoder has zero CF/CB separation, and only 3 dB from LF (or RF) to each back channel, it also sounds very spacious, noticeably more than QS, and surprisingly more than many discrete mixes. If SQ has a “sound”, that’s it.

But annoyingly, most of the logic-enhanced decoders threw it all away and sounded pretty heavily processed, similar to mistracking Dolby companding. The Sony/CBS “half-logic” decoder (which I owned) should have sounded spacious, but my memory was it was just sort of murky-sounding, and the logic sounded like it was always too late to do any good. My guess is the opto-coupling scheme used to accomplish the dynamic CF/CB blending was not very precise, and not very responsive.

In some important ways, slow dynamic decoding is worse than no enhancement at all. Most of annoying of all is the sense of some clown playing around with the mixer, fading sounds up and down for no good reason ... my biggest gripe with many commercial logic decoders. Once you start hearing that whoosh-up and whoosh-down sound, it’s impossible to ignore. (If I could hear it, I can only imagine how objectionable this level-shifting artifact would be for a recording engineer, who spend all day make tiny adjustments to balance and levels.)

Which is why I like very fast decoders that are engineered to be artifact-free. Sadly, very few people, including home theater reviewers, have ever heard one.

Done right, decoders of this type should sound very spacious and natural, almost like binaural recordings on headphones, but without the physical discomfort of phones and the weird front/back ambiguity of dummy-head recordings.

Without having heard your prototype, l imagine it sounds “transparent”, and is essentially artifact-free, thanks to digital lookahead and multiband processing. Modern direction-sensing algorithms also bypass the slowness and occasional glitching of earlier analog detection methods.

Most of all, with good ADCs and DACs, it should be spacious and wide-open sounding, which can “open-up” many squashed and compressed-sounding recordings. I particularly like that the type of reverb is instantly audible, just from its spatial signature, and that you can “hear into” the studio acoustic. This is what separates the best multichannel from the best 2-channel playback.

High-quality decoders are the *only* way to get this spatial quality of sound from the 99% of the existing 2-channel catalog. (Antiphase and 90-degree quadrature phase shifts are already present on stereo recordings. You just can’t hear it with 2 speakers or headphones. Trust me, I know. I’ve been designing speakers for a long time. But with a good decoder, you *can* hear what was there all along.)

I am also *very* confident that what you have is superior to the dull-sounding enhancement offered by modern home theater receivers.

 

[furui_suterioo]

For anyone out there developing quad stuff here are a couple of test file creation programs compiled for DOS. The first creates a tone that can rotate around any of the 3 Scheiber sphere axes (good for an Olson phase spin test). The second is a fairly comprehensive test tone generator for most quadraphonic encode methods.

I'm trying to run the tone generator but it wont stay open for more than a half second, tried on vdos, xp32, any thoughts?

 

[chucky3042]

Hi Malcolm, I’m so glad your subjective experience matches my memories of the SV hardware prototype. Although a static SQ decoder has zero CF/CB separation, and only 3 dB from LF (or RF) to each back channel, it also sounds very spacious, noticeably more than QS, and surprisingly more than many discrete mixes. If SQ has a “sound”, that’s it.

But annoyingly, most of the logic-enhanced decoders threw it all away and sounded pretty heavily processed, similar to mistracking Dolby companding. The Sony/CBS “half-logic” decoder (which I owned) should have sounded spacious, but my memory was it was just sort of murky-sounding, and the logic sounded like it was always too late to do any good. My guess is the opto-coupling scheme used to accomplish the dynamic CF/CB blending was not very precise, and not very responsive.

In some important ways, slow dynamic decoding is worse than no enhancement at all. Most of annoying of all is the sense of some clown playing around with the mixer, fading sounds up and down for no good reason ... my biggest gripe with many commercial logic decoders. Once you start hearing that whoosh-up and whoosh-down sound, it’s impossible to ignore. (If I could hear it, I can only imagine how objectionable this level-shifting artifact would be for a recording engineer, who spend all day make tiny adjustments to balance and levels.)

Which is why I like very fast decoders that are engineered to be artifact-free. Sadly, very few people, including home theater reviewers, have ever heard one.

Done right, decoders of this type should sound very spacious and natural, almost like binaural recordings on headphones, but without the physical discomfort of phones and the weird front/back ambiguity of dummy-head recordings.

Without having heard your prototype, l imagine it sounds “transparent”, and is essentially artifact-free, thanks to digital lookahead and multiband processing. Modern direction-sensing algorithms also bypass the slowness and occasional glitching of earlier analog detection methods.

Most of all, with good ADCs and DACs, it should be spacious and wide-open sounding, which can “open-up” many squashed and compressed-sounding recordings. I particularly like that the type of reverb is instantly audible, just from its spatial signature, and that you can “hear into” the studio acoustic. This is what separates the best multichannel from the best 2-channel playback.

High-quality decoders are the *only* way to get this spatial quality of sound from the 99% of the existing 2-channel catalog. (Antiphase and 90-degree quadrature phase shifts are already present on stereo recordings. You just can’t hear it with 2 speakers or headphones. Trust me, I know. I’ve been designing speakers for a long time. But with a good decoder, you *can* hear what was there all along.)

I am also *very* confident that what you have is superior to the dull-sounding enhancement offered by modern home theater receivers.

Yep we spent a lot of time working on the triband dual slope attack and decay characteristics, highly important!

 

[malcolmlear]

I'm trying to run the tone generator but it wont stay open for more than a half second, tried on vdos, xp32, any thoughts?

Hi, well it wont stay open for any great deal of time, just sufficient to create the test tone file.

 

[furui_suterioo]

Hi, well it wont stay open for any great deal of time, just sufficient to create the test tone file.

I dont have much experience with dos, I installed Vdos, put the testtonewavegen in the Vdos directory and it shows TESTTO~1, and it doesn't seem to want to execute command starting in TestToneWaveGen, I probably just don't have the right syntax to start the program, where does it store the output file?

 

[BJMarsh]

Wanted to share some of my dabbling with my software SQ project. As a side project I coded an SQ 10/40 blend algorithm. I was pleasantly surprised to hear some really nice space on Ten Years After’s A Space in Time. Nice surround but not a lot of separation. I did need to invert the right channel because the vocals were in the back! The SV algorithm also required this fix as well. The source was the culprit because my test signals confirm that my basic decoding is correct. Unless the intent was to put the vocals in the rear. Unlikely.
The reason why I tried the 10/40 blend was to see what DTS-Neo 6-Music would do with it. It’s supposed to be a multiband logic steering algorithm. Well there’s no such thing as a free lunch!

 

[BJMarsh]

For Lynn & Malcom... The CF->RB->CB->LB->CF rotation works but my look ahead leaves a something to desired. Root cause is likely the way I compute the averages for the comparators. I need to figure out how to attack quickly / decay slowly without switching (?) distortion... Never say die! As it stands now, it is far superior to the Sony “Full Wave” decoder from the 1970’s. This has been fun to do. This is my last hurdle before I tackle multiband.

What the heck did “Full Wave” mean??? Full price? That decoder made me turn away from SQ completely. I didn’t buy another SQ record again until two months ago. Damn shame because the SV has been a revaluation. Thanks.

 

[malcolmlear]

I dont have much experience with dos, I installed Vdos, put the testtonewavegen in the Vdos directory and it shows TESTTO~1, and it doesn't seem to want to execute command starting in TestToneWaveGen, I probably just don't have the right syntax to start the program, where does it store the output file?

Hi
You shouldnt need to install anything, just run from a command prompt or create a bat file. The files are created in the same directory as the executable. I'm using Windows 7, but cant see why it wont run in 8 or 10.

 

[furui_suterioo]

Hi
You shouldnt need to install anything, just run from a command prompt or create a bat file. The files are created in the same directory as the executable. I'm using Windows 7, but cant see why it wont run in 8 or 10.

I've got the TestToneWaveGen\TestToneWaveGen.exe in root C:\ folder. I entered
C:\TestToneWaveGen\TestToneWaveGen QS 96000 24 8 10 1000
Into cmd, it responds with
(c) Malcom Lear 2020
Version 1.04
but I dont see any new files in the directory.

 

[BJMarsh]

Lynn,

I have a quick question regarding your SQ Shadow Vector patent. If I am reading it correctly, then VGA150, 152, 166 and 168 should have a gain between 0 and 1.5? This would result in a final gain of 0.5 to 2.0 ( -3db to + 3db) when summed with the fixed gain amplifiers for LF, RF, LB and RB. I have interpreted this correctly?

Hope all is well in CO. Best regards,

Blaine

 

[Lynn Olson]

Colorado is just coming off it’s second wave, but our little family is still sheltering-in-place until the vaccine arrives. My best guess is some time in the 1st quarter of 2021, which puts us at the halfway mark about now. Still the same routine of multiple daily dog-walks, weekly groceries that get delivered at the entrance to the garage, and weekly short drives to keep the cars going. Lots of British TV in the evenings, streaming through Amazon, Britbox, and Acorn. Can’t stand US TV, especially the grotesque reality shows.

It’s weird how a lot of the younger people seem to think this is over. It ain’t, not until a fair number of people are vaccinated and/or get mild cases so we can get herd immunity. But the mortality rate for our demographic is 2%, and the hospitalization rate is 10%, so we’re playing it as safe as we can. Don’t want to mess with this thing, not at our age.

But the mind is working OK, and I‘ve started reading science fiction again, starting with Iain M Banks and the Culture series. Now I’m working through Asimov. Maybe Cordwainer Smith after that. Visionary science fiction is a good mental alternative to the daily horror coming out of the TV.

I am glad to see how younger folk are re-discovering quadraphonic sound. Digital processing has many advantages compared to the insane complexity of doing it all in analog. Doing it the hard way reminds me of color TV in the all-vacuum-tube era ... 30 tubes or more, not counting the expensive 21” or 25” crt. Not cheap, and setup could be pretty involved, with initial degaussing, purity-magnet setup, and then multiple passes of convergence setup (yes, I’ve done it).

I was thinking the other day about the suitability of SQ decoding as an enhancement of stereo recordings, versus QS decoding. CB is the same, of course, but the treatment of 90-degree phase-shifted content is quite different. In QS playback, it is overhead between all four speakers, while it is LB or RB in SQ, depending on which channel has the advanced phase.

With spaced microphones or plate reverb, you automatically get random-phase reverb, which then illuminates the entire surface of the Scheiber sphere. This illumination of the Scheiber surface is only moderately audible with 2-speaker playback, but is very clear and obvious with either a static SQ or QS decoder, or a high-quality SQ or QS dynamic decoder. I had been thinking that QS might have better quality for stereo content, but I’m re-considering that. In SQ, the difference between LB and RB is a matter of which channel has the leading phase, and, well, transients recorded with spaced microphones does result in the appropriate channel leading. So maybe SQ playback of stereo recordings might be best after all.

(An aside: it’s usually thought that a 90-degree phase spread smears the sound between 2 speakers in stereo playback, but that’s only true for smooth sounds without sharp transients. If transients are present, they tend to appear from the speaker with the leading phase, which is why spaced-microphone recordings have better image quality than you might expect.)

SQ’s strength is preservation of full separation across the frontal arc, something it shares with stereo, and is quite different than QS in that respect. Hmm, food for thought.

Oh yes, your question. Hmm, not sure. Been ages since I’ve done the math on the VGA section. The goal is for the overall decoder to have exactly the same net gain as a static, unmoving SQ decoder, and for each dynamic channel (vector) to follow a panpot law as it moves around in various directions. The patent could have been a little clearer on that point.

The primary SV goal, not really mentioned in the patent, is for random-phase energy on the surface of the Schieber sphere to be unaffected by the decoder dynamics, no matter what it is doing at the moment. This is where other dynamic decoders tended to fall down ... they modulated the random-phase energy, and this is very audible and objectionable. I tried to explain this concept to the CBS team, and I think I just offended them. And I never did set up a line of communication with the Tate DES team, so I have little idea of their goals.

This constant-power feature would be especially important for a multiband decoder, otherwise some very unnatural colorations could creep in. Perhaps that’s the problem with the Dolby ProLogic II and DTS Neo decoders, which seem to have transient colorations that come and go.

Huh. The box under my name says I joined the forum ten years ago. How time flies.

 

[malcolmlear]

I've got the TestToneWaveGen\TestToneWaveGen.exe in root C:\ folder. I entered
C:\TestToneWaveGen\TestToneWaveGen QS 96000 24 8 10 1000
Into cmd, it responds with
(c) Malcom Lear 2020
Version 1.04
but I dont see any new files in the directory.

Hi there,
Not sure what the problem is. I tried the same parameters as you and it produced the attached file. It was compiled on a 32 bit windows machine, but I would have thought it would run fine on 64 bit.
Cheers,
Malcolm

 

[furui_suterioo]

Hi there,
Not sure what the problem is. I tried the same parameters as you and it produced the attached file. It was compiled on a 32 bit windows machine, but I would have thought it would run fine on 64 bit.
Cheers,
Malcolm

I'm now reading reading that 64bit windows 10 doesn't like dos programs, so I'll try in a 32bit vm. Thanks for the file though.

 

[Owen Smith]

There's a world of difference between a 32 bit command line Windows program, and a DOS program (which is 16 bit). It's been a l-o-n-g time since Microsoft supplied a compiler that can build DOS programs. 64 bit Windows 10 is perfectly happy with 32 bit command line programs.

 

[furui_suterioo]

There's a world of difference between a 32 bit command line Windows program, and a DOS program (which is 16 bit). It's been a l-o-n-g time since Microsoft supplied a compiler that can build DOS programs. 64 bit Windows 10 is perfectly happy with 32 bit command line programs.

Its working great on windows xp in virtualbox, but for some reason I get no output file in windows 10, useful program for comparing decoders. Thank you

 

[BJMarsh]

I have my algorithm somewhat sorted out. My attack/decay isn’t there yet and I haven’t heard the SM2 or Malcom’s decode yet so I have no idea whether my decoder is any good. I can say there are times when the separation is impressive. What is a constant is the ambience and surround. It sounds like Lynn’s description. I am looking for folks to critique the algorithm. I can supply some samples.

 

[BJMarsh]

If anyone has samples of SM2 or Malcom’s SQ decoder I would love to hear those.