A hundred years ago,
or so it seems, when I was just kid in high school, I picked up the latest issue of Popular Electronics November 1970. In that issue was a column by some new guy named J. Gordon Holt (who later founded Stereophile magazine). It was titled “Hafler vs Scheiber: 4 channels on discs”. It was really intriguing to me and the fairly simple differencing rear speaker hook up was too easy not to try. So I hooked up a spare 12” naked speaker driver a friend gave me from his Pizza Hut PA system. KLH model 55 in front, Pizzaphonic sound behind me, Walter Carlos’ “Switched On Bach” (stereo of course) on the turntable.
Some things in life just click, you know? I fell in love with the additional sound from the rear. From that more than humble start I’ve always had some form of surround sound. So much hardware & software has come and gone since then but I still love music in surround. And I still have that issue of Popular Electronics.
Much more recently I happened across the QQ forum thread on decoding SQ on a PC. It was every bit as thought provoking as that first article I read about 4 channel sound. Right now I’d like to give a big THANK YOU to one and all who contribute such great stuff to the QQ forums.
Part of that note thread touched on QS decoding and the subject has risen elsewhere. Actually I don’t have much interest (right now anyway) in decoding either SQ or QS on my PC. But I do have a very strong desire for synthesizing surround from stereo. And over the years I’ve been convinced that the QS format is the best place to start. One of the comments caught my attention “it seems so that the QS matrix is not as sensible as SQ.“ This and other attempts made to decode QS on a PC creates an implication that QS is some how more difficult to decode on a PC when drawing on prior experience of accomplishing it with SQ. This ran counter to everything I knew in regard to analog comparisons. That is, when designing a basic decoding circuit or applying any separation enhancement (such as Tate DES) to QS it is always simpler than doing the same for SQ. There must be a way to do it reasonably on a PC.
So what I’ve come up with & want to share is NOT a method for decoding QS on a PC but an approach to synthesizing four or more high separation channels from stereo ( S2S, Stereo to Surround) based on the QS decoding format. I will leave it to others to adapt my tricks for the intent of proper decoding of QS. My goal was to create surround sound that that at least equals that from a QS VarioMatrix or Fosgate 101 and exceeds the rather weak wrap around sound from PL 2. I have accomplished this.
What it does:
1, This method can create very subtle to extreme wrap around sound. It achieves that Holy Grail of stereo to quad: a stereo sound panned left to right will (if you desire) start in left back only, move smoothly to left front, center front, etc. and end up in the right back speaker.
2, It is tonally very neutral so a LFE is definitely doable but strictly optional.
3, Center front phantom imaging is very good, providing it exists in the original. Heavily chorused or FX’ed vocals could be positioned anywhere. A natural center front vocal will image beautifully. A dedicated center front channel is also therefore optional.
4, This approach is very flexible in terms of pre & post decoding editing & manipulation. You can choose wrap around effects differently for different songs in a compilation or even change decoding for an intended purpose within a song, since none of this is real time.
What it isn’t:
1, Fast & easy. Nothing is easier than just plopping a disc in and turning on your favorite decoder. Using scripts, batch processing & doing this on not just one but two apps of Adobe Audition 1.5 simultaneously I can go from unripped CD to DTS files ready to burn in about an hour. Or of course if it’s a special project you can put as much time & creativity as you want into it.
2, It is not VarioMatrix. Sansui used a true variable matrix circuit that adjusted itself based on the input signal. Doing it on a PC requires first simple basic QS decoding to be done then separation enhancement follows.
3, It does not involve a bunch of sum/difference signals or any 90 deg phase shifts.
The key to all this is apparent when you consider the crosstalk that occurs in simple QS decoding. I know I’m stating the obvious for a lot of people here but here are the decoding results for a corner signal ( say, left front) when decoded QS without VarioMatrix:
Left Front = equals full level
Right back= zero output
Left back & right front carry the crosstalk equally at only 3dB less than Left front.
So it is obvious that in PC decoding attention needs to be focused on reducing crosstalk on a diagonal basis. Adobe Audition has the perfect tool for this, the Center Channel Eliminator. To the best of my knowledge this CCE exists only in Adobe Audition & nothing quite like it in competitor’s software. Also it is only in the 1.5 version of AA. As a sidebar, another difference to think about between my approach and VarioMatrix is the latter does it’s job of creating corner speaker feeds by using essentially center channel signals, that is, L+R, L-R, Left only, Right only. It offers very good center channel performance due to this but it is a little bit of doing things indirectly when the desire is to make corner speaker feeds. My approach decodes the corner speaker feeds first and then applies crosstalk reduction that takes place simultaneously in both front/back & left/right areas.
So here it is as brief as I can say it:
Decode & create two individual stereo pairs in AA 1.5 that represent QS Lf/Rs and Rf/Ls. Apply AA CCE to each pair to reduce common crosstalk. There you have it.
With everyone’s patience I am going to break this into another part & give the details later. My two little index fingers are just bloody stubs now from all this time at the keyboard. I know some of you will immediately begin testing this out & proceeding ahead. There are roadblocks to this process not so apparent & shortcuts to be used that I’ve tweaked over the last month or so I’ve been doing this. Give me a day or so & I’ll be back with the real How To.
R. Scott Varner
“Nothing stops the Creative Spirit”, M. Jittlov, ‘89
or so it seems, when I was just kid in high school, I picked up the latest issue of Popular Electronics November 1970. In that issue was a column by some new guy named J. Gordon Holt (who later founded Stereophile magazine). It was titled “Hafler vs Scheiber: 4 channels on discs”. It was really intriguing to me and the fairly simple differencing rear speaker hook up was too easy not to try. So I hooked up a spare 12” naked speaker driver a friend gave me from his Pizza Hut PA system. KLH model 55 in front, Pizzaphonic sound behind me, Walter Carlos’ “Switched On Bach” (stereo of course) on the turntable.
Some things in life just click, you know? I fell in love with the additional sound from the rear. From that more than humble start I’ve always had some form of surround sound. So much hardware & software has come and gone since then but I still love music in surround. And I still have that issue of Popular Electronics.
Much more recently I happened across the QQ forum thread on decoding SQ on a PC. It was every bit as thought provoking as that first article I read about 4 channel sound. Right now I’d like to give a big THANK YOU to one and all who contribute such great stuff to the QQ forums.
Part of that note thread touched on QS decoding and the subject has risen elsewhere. Actually I don’t have much interest (right now anyway) in decoding either SQ or QS on my PC. But I do have a very strong desire for synthesizing surround from stereo. And over the years I’ve been convinced that the QS format is the best place to start. One of the comments caught my attention “it seems so that the QS matrix is not as sensible as SQ.“ This and other attempts made to decode QS on a PC creates an implication that QS is some how more difficult to decode on a PC when drawing on prior experience of accomplishing it with SQ. This ran counter to everything I knew in regard to analog comparisons. That is, when designing a basic decoding circuit or applying any separation enhancement (such as Tate DES) to QS it is always simpler than doing the same for SQ. There must be a way to do it reasonably on a PC.
So what I’ve come up with & want to share is NOT a method for decoding QS on a PC but an approach to synthesizing four or more high separation channels from stereo ( S2S, Stereo to Surround) based on the QS decoding format. I will leave it to others to adapt my tricks for the intent of proper decoding of QS. My goal was to create surround sound that that at least equals that from a QS VarioMatrix or Fosgate 101 and exceeds the rather weak wrap around sound from PL 2. I have accomplished this.
What it does:
1, This method can create very subtle to extreme wrap around sound. It achieves that Holy Grail of stereo to quad: a stereo sound panned left to right will (if you desire) start in left back only, move smoothly to left front, center front, etc. and end up in the right back speaker.
2, It is tonally very neutral so a LFE is definitely doable but strictly optional.
3, Center front phantom imaging is very good, providing it exists in the original. Heavily chorused or FX’ed vocals could be positioned anywhere. A natural center front vocal will image beautifully. A dedicated center front channel is also therefore optional.
4, This approach is very flexible in terms of pre & post decoding editing & manipulation. You can choose wrap around effects differently for different songs in a compilation or even change decoding for an intended purpose within a song, since none of this is real time.
What it isn’t:
1, Fast & easy. Nothing is easier than just plopping a disc in and turning on your favorite decoder. Using scripts, batch processing & doing this on not just one but two apps of Adobe Audition 1.5 simultaneously I can go from unripped CD to DTS files ready to burn in about an hour. Or of course if it’s a special project you can put as much time & creativity as you want into it.
2, It is not VarioMatrix. Sansui used a true variable matrix circuit that adjusted itself based on the input signal. Doing it on a PC requires first simple basic QS decoding to be done then separation enhancement follows.
3, It does not involve a bunch of sum/difference signals or any 90 deg phase shifts.
The key to all this is apparent when you consider the crosstalk that occurs in simple QS decoding. I know I’m stating the obvious for a lot of people here but here are the decoding results for a corner signal ( say, left front) when decoded QS without VarioMatrix:
Left Front = equals full level
Right back= zero output
Left back & right front carry the crosstalk equally at only 3dB less than Left front.
So it is obvious that in PC decoding attention needs to be focused on reducing crosstalk on a diagonal basis. Adobe Audition has the perfect tool for this, the Center Channel Eliminator. To the best of my knowledge this CCE exists only in Adobe Audition & nothing quite like it in competitor’s software. Also it is only in the 1.5 version of AA. As a sidebar, another difference to think about between my approach and VarioMatrix is the latter does it’s job of creating corner speaker feeds by using essentially center channel signals, that is, L+R, L-R, Left only, Right only. It offers very good center channel performance due to this but it is a little bit of doing things indirectly when the desire is to make corner speaker feeds. My approach decodes the corner speaker feeds first and then applies crosstalk reduction that takes place simultaneously in both front/back & left/right areas.
So here it is as brief as I can say it:
Decode & create two individual stereo pairs in AA 1.5 that represent QS Lf/Rs and Rf/Ls. Apply AA CCE to each pair to reduce common crosstalk. There you have it.
With everyone’s patience I am going to break this into another part & give the details later. My two little index fingers are just bloody stubs now from all this time at the keyboard. I know some of you will immediately begin testing this out & proceeding ahead. There are roadblocks to this process not so apparent & shortcuts to be used that I’ve tweaked over the last month or so I’ve been doing this. Give me a day or so & I’ll be back with the real How To.
R. Scott Varner
“Nothing stops the Creative Spirit”, M. Jittlov, ‘89
