Involve Encode/Decode

[Sonik Wiz]

The news of the upcoming Surround Master V2 has got me really revved. I've been re-reading many of the posts here & stumbled across some new stuff I've hadn't seen before such as this:

https://www.hifizine.com/2013/03/th...ew-with-charlie-van-dongen-and-rob-mackinlay/

I have also checked out the Involve patent on their encode system that I never had much interest in before. It is pretty much the Sansui QS encode with drastically reduced Left/Right blending:

L=FL+kFR+jRL+jkRR
R=FR+kFL−jRR−jkRL

wherein k denotes a transformation or matrix constant having a value approximately 0.207 and j denotes a 90 degree phase shift.

This .207 blend is about 13dB compared to QS ,414 about 7dB. Much is said in the patent that a listener can not hear over 12dB separation in a regular listening environment. I interpret the attempt is to make surround encoded material closer to a regular non-encoded stereo source.

Testing with the full decoder described in PCT/AU2010/001666 resulted in 12 db separation in the 4 surround output signals. During the testing listeners could not hear the difference between the 12 db matrix and the 40 db matrix or discrete surround sound. In addition listeners also could not hear the difference between the encoded surround stereo and normal stereo.

It still seems an odd choice not to have complimentary encode/decode. If there is a demonstrable separation of 12dB, what position/direction is this? Seems like there is the potential for wide L/R at the expense of F/B.

Is there any commercial music released in the Involve encode? Beside that odd situation I've read about here on QQ where you buy an album & get a decoder along with it.

It also makes me interested if the SM V1 is outputting maximum separation or just 12dB?

Hope to hear from Chucky on the decision making for non-matching encode/decode.

 

[chucky3042]

Hello Sonik Wiz

Great questions and thanks for starting the thread......one of my favourite surroundy topics!

Yes that article on the electrostatics was me!!

There are two patents we made covering the encode, I think you are referring to our "dumb encode" patent where upon a bunch of testing with test monkeys we found that there was magic around the 12 dB separation figure (also lines up with the Haas precedence effect 12 dB number, that we use for our SST sit anywhere patent).

The question arose as to how many dB separation is required where the listener could no longer perceive an improvement in separation, we set up 10 listeners and some test jigs that deliberately created crosstalk and the controversial result came in ...........roll drums...................................12dB! On double blind . no bullshit testing no one could pick any improvement beyond that number. I know this is opposite common audiophile "wisdom" in a world where CD's achieve 100 dB but it goes a long way to explaining why some old phono cartridges such as the empire (with around 12 db separation) sounded fine. I also know all the QQ types love their 40 dB number. The point of that patent was the discovery that with our Involve matrix encoder set to the 0.21 parameter as you point out you end up in an interesting area of space where both the stereo encoded signal had 12 dB separation AND the resultant surround decode had 12 dB separation IN ALL DIRECTIONS. So in terms of a listening experience both the stereo and surround listeners think they are getting full normal reproduction - yes I know its controversial - blah, blah, blah.

The dumb patent was needed to lay claim to that bit of IP. This was not used for any of our products. Please note I have attached two channel separation tables below, the first one is our involve decoder decoding from a straight flat QS matrix and as you can see we achieve 35 - 40 dB in all directions - YAY!

Below the QS separation table is another not as impressive number table showing the STATIC results of the INTELLIGENT INVOLVE ENCODE/ DECODE matrix. As you can see in two directions it got down as low as the mythical 12 dB number. I have also attached a pdf of our "smart encode" matrix patent and what needs to be understood is that the encoder is now a full variable parameter matrix that can vary the 0.21 parameter from any number from 0 to 0.414 number (in practice we restrict the lower point to a secret value). The actual choice of the matrix parameter is decided by the actual musical content - for a heavily stereo content signal with little surround content the parameter used will be a lower number - putting emphasis of the stereo. For music with heavy surround content the emphasis on this parameter is adjusted to a number higher than 0.21.

It needs to be remembered that this parameter is dynamically changing all the time to match the music content - more importantly it is a triband parameter so one band might have a high number but another band will have a low number but flip around a few milliseconds later. The net result is to both the stereo and surround listener they actually perceive full separation.

For those who want to play around with this we have made evaluation boards of both the encoder and decoder, see attached- they are quite low priced!

Hope this clarifies the mud - sounds good!

Re music with Involve encode - Dave the Bitch (Overture) will comment next...................

Regards

Chucky

 

[chucky3042]

Err....here is the separation tables attachment

 

[Overture]

Hi all,

In answer to your question - the first album released in our format is now available on Itunes / Amazon / ETC in digital form.
There is also a limited Vinyl box set which includes a custom decoder we designed for the project. It's Suzanne Ciani's LIVE quadraphonic album, and we have been working with her and the producer KamranV on this project all year.

https://itunes.apple.com/us/album/live-quadraphonic/1380954771
https://www.amazon.com/Live-Quadraphonic-Suzanne-Ciani/dp/B07CVK7SB8

There's also a write up about it in Wired among other places.

https://www.wired.com/story/suzanne-ciani-live-quadraphonic/


So, you know. There's that.

 

[chucky3042]

So just to clarify a point. For the music with actual surround content the encode parameter stays pretty close to the 0.414 number but drops lower in passages where it is more stereo based. The net result is undetectable to the listener. The DYNAMIC separation will be close to the 35 - 40dB range, with no compression of the stereo encode source as happens with QS material. It works well also because of the high sensitivity of the phase detectors in the INVOLVE decoder.

Regards

Chucky

 

[Sonik Wiz]

Chucky
Thank you for your most generous reply!
About a hundred years ago I considered a design using a variable encoder & fixed coefficient matrix decoder. Intuitively it seemed something could be done better utilizing the high separation between QS decoded diagonal channels. Then I gave up realizing that I could probably improve corner separation but not do anything for F/B or L/R. Of course your overall design concept is so much more advanced.

RE: 12 dB
The NRDC Ambisonic team did testing like this and also found 12dB separation as a landmark specification. They spoke in a somewhat broader sense saying that 12dB channel separation is the minimum (my emphasis) necessary for good directional location. Lucky for them this is what the three channel discrete Type B Format offers, with diagonal opposite corners silent.

I will look over the charts & data attachments you posted more closely later. Thanks! I can see some variables that might affect the test results showing 12dB as the magic point. What sort of music was used? High separation pop or classical? Speakers being bi-polar stat's or direct firing? Reveberant room acoustics could certainly affect it. And last but not least we're the test monkey's experienced Quadraphiles?

In the 80's I played PF's Money for a friends wife decoded SQ Tate. Everyone knows this song & how could it not impress her? Afterwards she said" it sounds just like stereo, only louder". Some people hear things and others don't. Yup.

I don't mean to fixate on the 12dB spec too much. Because it's more amazing you've taken a concept almost 50 years old & brought it to life again. Even in the 70"s had they tried your approach of variable encode/decode doing it in all analog would have probably done more harm than benefit.

Good on ya!

 

[Sonik Wiz]

Hi all,

In answer to your question - the first album released in our format is now available on Itunes / Amazon / ETC in digital form.
There is also a limited Vinyl box set which includes a custom decoder we designed for the project. It's Suzanne Ciani's LIVE quadraphonic album, and we have been working with her and the producer KamranV on this project all year.

https://itunes.apple.com/us/album/live-quadraphonic/1380954771
https://www.amazon.com/Live-Quadraphonic-Suzanne-Ciani/dp/B07CVK7SB8

There's also a write up about it in Wired among other places.

https://www.wired.com/story/suzanne-ciani-live-quadraphonic/


So, you know. There's that.

Hello Mr Overture!

Alas I do not do either itunes or amazon or anything MP3 if avoidable. Should this become avaialbe on HD Tracks or the such I would immediately buy it. I wish you success and good sales results from this!

 

[Overture]

Hello Mr Overture!

Alas I do not do either itunes or amazon or anything MP3 if avoidable. Should this become avaialbe on HD Tracks or the such I would immediately buy it. I wish you success and good sales results from this!

I tend to agree. AFAIK there will be HD digital versions available, when and where is a question yet to be answered, but I'm totally on board.

Cheers.

 

[chucky3042]

Chucky
Thank you for your most generous reply!
About a hundred years ago I considered a design using a variable encoder & fixed coefficient matrix decoder. Intuitively it seemed something could be done better utilizing the high separation between QS decoded diagonal channels. Then I gave up realizing that I could probably improve corner separation but not do anything for F/B or L/R. Of course your overall design concept is so much more advanced.

RE: 12 dB
The NRDC Ambisonic team did testing like this and also found 12dB separation as a landmark specification. They spoke in a somewhat broader sense saying that 12dB channel separation is the minimum (my emphasis) necessary for good directional location. Lucky for them this is what the three channel discrete Type B Format offers, with diagonal opposite corners silent.

I will look over the charts & data attachments you posted more closely later. Thanks! I can see some variables that might affect the test results showing 12dB as the magic point. What sort of music was used? High separation pop or classical? Speakers being bi-polar stat's or direct firing? Reveberant room acoustics could certainly affect it. And last but not least we're the test monkey's experienced Quadraphiles?

In the 80's I played PF's Money for a friends wife decoded SQ Tate. Everyone knows this song & how could it not impress her? Afterwards she said" it sounds just like stereo, only louder". Some people hear things and others don't. Yup.

I don't mean to fixate on the 12dB spec too much. Because it's more amazing you've taken a concept almost 50 years old & brought it to life again. Even in the 70"s had they tried your approach of variable encode/decode doing it in all analog would have probably done more harm than benefit.

Good on ya!

Hi Sonik Wiz

We conducted our tests with a mix of test monkeys with 60% being experienced audio nuts and a variety of music including pink noise and test tones. We included music with high ping pong style separation and even spread stuff. More popish/ rock/hill-billy than classical music (we are low class types). All speakers used were electrostatic dipoles (we are one eyed on that!)

You need to consider the following points:

1 The minimum instantaneous separation for both the stereo and resultant surround will be 12 dB, for the bulk of the time it is in fact way higher towards 40 db. So the minimum is inaudible in terms of crosstalk.

2 For complex musical passages with lots of surround the separation will be 35 to 40 db

3 For music content with minimal surround the stereo is emphasised so it get a full 12 db separation.

4 Its all done in 3 bands and so the listener will be more unaware of the background wiggles

5 Music is not static......it is dynamic so static separation numbers are fairly meaningless.

Regards

Chucky

 

[Simon A]

We conducted our tests with a variety of music including pink noise

The song used was Comfortably Hum.

We included music with high ping pong style separation...

Also known as Forrest Gump style separation...

 

[DuncanS]

Could have scared the audience with some White Noise...…...the 1969 variant that is!

What does the final track Black Mass: An Electric Storm In Hell sound like decoded by the Surround Master?

 

[Sonik Wiz]

.

 

[Sonik Wiz]

I finally had a chance to give a close look at the Involve smart encoder. I can see why you are so proud of it. Looking at the pic of the demo encoder board what I can’t see is how you got all of that on to what seems a pretty small board. Wow!

Figure 8 is missing.

I see that the patent now lists the minimum coefficient blend at .1 (-20dB) and max at .414 (-7.6 dB). So this amount of control is actually centered on a 12.4 dB range? It’s easy to understand in the of high L/R separation it would be minimum blend and high for front or back signals. It seems that a Left front input would be read has having both high L/R and F/B separation. This would put the blend control signal at an arbitrary mid position. What would the blend coefficient be in that situation?

On the Involve website there are two different encoders shown. One has a blue LCD window on the left side & the other does not. Also different control lay out. Both are labeled IM-ED1. What are the differences? I also see two knobs for Output Mix. One is labeled Encoder & the other labeled Ambient. These are not covered in the patents. What cool things do these do?

And when the encode is set to fixed what blend is used for that?

 

[chucky3042]

I finally had a chance to give a close look at the Involve smart encoder. I can see why you are so proud of it. Looking at the pic of the demo encoder board what I can’t see is how you got all of that on to what seems a pretty small board. Wow!

Figure 8 is missing.

I see that the patent now lists the minimum coefficient blend at .1 (-20dB) and max at .414 (-7.6 dB). So this amount of control is actually centered on a 12.4 dB range? It’s easy to understand in the of high L/R separation it would be minimum blend and high for front or back signals. It seems that a Left front input would be read has having both high L/R and F/B separation. This would put the blend control signal at an arbitrary mid position. What would the blend coefficient be in that situation?

On the Involve website there are two different encoders shown. One has a blue LCD window on the left side & the other does not. Also different control lay out. Both are labeled IM-ED1. What are the differences? I also see two knobs for Output Mix. One is labeled Encoder & the other labeled Ambient. These are not covered in the patents. What cool things do these do?

And when the encode is set to fixed what blend is used for that?

Hi Sonik Wiz

Not sure what you mean about figure 8 is missing.....?

Actually we center the coefficient closer to 0.27 (from memory), as in all things we do we extensively listen to it and often we go against theory. For the encode signal just in the Left front as you describe the system would deem a low surround - high stereo content and as such the coefficient would be a low number - probably around 0.15 (as we actually restrict it).

Re the pro studio encoders - yes we had 2 versions. The first one with the blue LCD also had a pre amp built it - so it was a "one stop shop" but the thing had a RRP of $2990. We sold a few in Australia but we redesigned it to be more dedicated to studio and lowered the price to below $2 K . We have not done a production run due to internal finances being too tight.

The two knobs on the front panel " Encoder" and "Ambient" are perhaps bad terms but they actually facilitate the possibility of mixing two formats (in varying ratios) - so Dolby for example would be denoted as the ambient.

On the encoder evaluation module with the switch set to fixed the coefficient is set to 0.25. I think inadvertently (by lack of communication on our part) the Suzanne Ciani recording was made in this mode.....not really totally sure.

Regards

Chucky

 

[chucky3042]

Dear All

Just as we are discussing technical stuff, I note that from big Max's tests on a QS stereo input with a surround coefficient of 0.41 we get the following separation:

When you compare it to the good ol Sansui QSD 1 (with stuffed right hand output buffers)

You will note that we get 10 - 15 dB more separation. This is partly due to doing all the maths in the digital domain without the tolerance issues of analogue circuitry. In addition we do tons of other tricky psychoacoustic and vector based stuff plus interesting audio "pick and place" tricks. But what it means is that we can get more separation even with lower values of the matrix "constant".

Now don't go all nuts that somehow we are running a low separation system as the Surround Master as it is permanently set to maximum surround decode of 0.41 coefficient on all 3 bands. All the tricks are happening on our smart encode matrix - when there is surround you can bet the constant will be up around the 0.41 number on that particular frequency band.

Just sayin

 

[~dave~~wave~]

...All the tricks are happening on our smart encode matrix...

Appreciate the technical discussions, even though most of it's over my head.

What I can say is how impressed I am with the ENCODE of the live Suzanne Ciani.
It sounds noticeably more discrete in DoProLogII Music than SQ encoded stereo tracks I've tried.
I've listened to it several times now.

It's fun to switch back and forth between stereo & quad decode.
The stereo image seems extra-wide, perhaps from the processing?
Decoded, the rear sound field and surround panning are striking.

Proof of concept for the encoder, IMHO.
Congratulations to the inventors.

Do any of you folks who actually own a Surround Master have comments on how this decodes on your unit?

Check it out:
http://www.quadraphonicquad.com/for...format-suzanne-ciani-live-quadraphonic.25225/

 

[Sonik Wiz]

I'm right there with Dave. Enjoyed it multiple times. Great music, great soundfield. I have not listened to it stereo maybe tonight. I have listened to it with SpecWeb at default settings and the Involve encode works very good with that as well. Compared to DPL II I'd say the pans & locations are much the same but DPL II seems more speaker centric while SpecWeb seems to be simply larger. Yeah, hard to verbalize...

Looking forward to buying or DL'ing a legit high quality source & play back on new SM V2!