Stereo Pre-Synthesis

[Sonik Wiz]

More fully: manipulating in 2 ch stereo on a PC to produce a desirable outcome when playing through a matrix decoder such as QS Variomatrix, etc.

For the most part when decoding stereo into surround I have had two ways of doing it. One is most casual such as simply playing through my Sansui QSD-1000, Fosgate Tate 101A, or DPL II on my Anthem AVM30. Otherwise I use the upmixing method I described several years ago on this forum for the really important music I wanted to enjoy in the best possible way. The latter gives excellent results but is very labor intensive. For oh maybe the last six months I have been sampling an intermediate method which is adjusting phase balance , EQ, panning, etc. on Adobe Audition prior to playing back on a decoder. It’s common knowledge that stereo into surround is quite variable & mostly it relies on how close the two channel source comes to matching the requirements of the decoder matrix. Controlling it this way puts a lot of power into making the original source do that. Even in the age of the Surround Master this is still valid since the SM does not have a synth or hall function as on old VarioMatrix decoders that can be quite useful. This method also works great when using SpecWeb at default settings. Also EQ can be applied. Doing out of phase blending can reduce the bass & that can be compensated for. I tend to listen to a lot of J Pop/Rock that is often pretty bright & some graphic EQ takes that edge off better than simply adjusting bass/treble controls.

The most useful tools in Audition for this is Channel Mixer and Pan/Expand (process) under Stereo Imagery. Channel mix can be used to create in phase or opposite phase mixing. Set on Wide Field it produces close to a -10dB opposite phase blend that idens the image in stereo & goes a long way to opening up the sound field when decoded. It is easy to make a preset that matches Sansui’s synth mode:

In this case a left only guitar in stereo will play back in left back with sharp directionality in QS , DPL II when adjusted right and I’m sure the SM too. A guitar panned left to right will make that beautiful 270 deg horseshoe enhancement that always adds so much to pop & rock.

The other tool I use is the Pan/Expand (process). I have never seen any comments on this tool mentioned before on the QQ forum so I wonder if anyone knows much about it. It widens the stereo field and provides out of phase blending but works much different than the channel mixer. On one hand it does dynamically make the dominant channel in the stereo mix even louder. It also creates out of phase blending towards the sub-dominant channel. Because it is dynamic it has most expansion when a single channel is present and zero action when the levels match as in a mono mix. This means you get opposite phase enhancement without a deficit of center front vocals or bass as you do when using channel mixer. A preset of 200 matches almost exactly the channel mixer Wide Field at -10dB blend and set at 250 you get the -7dB blend:

Neither is perfect I find one of the two might work better for a particular song or album. Pan/Expand will increase the signal levels so I usually Normalize at -6dB before using and eventually when all is done it’s my habit to normalize each song to -1dB. Maybe I state the obvious but you can also use these tools to mix in phase & bring the sound more up front. With the Pan/Expand you can create variable blending over selected time frame. That is you could choose a song that has a subtle mellow opening adjusting for full mono blend and gradually open up to original stereo & expand to a synth setting. This worked out very nice for G n’ R’s November Rain.

Also in Pan/Expand you could take (as an example) a solo lead guitar that is just in left channel & bounce it left/right in pan. Doing that and adding out of phase blend you now have a guitar that will bounce between left and right back or if slower it will pan around 270 deg.:

As mentioned before this is also a good time to do EQ. I like to tweak thing so my bass/treble can be set at zero. I’m not OCD about that; it’s just that it’s so easy to get more accurate tone control beforehand.

If this seems like a lot of trouble I can only say it’s much easier than my full blown method that takes hours to do on an album. My Delta 1010 (not LT) SPDIF goes straight to my Anthem AVM30. I can sit in the sweet spot; look at Audition on an eight foot screen and with a wireless mouse & keyboard I can hear exactly how it’s going to play back. I think it is pretty easy to make quick decisions listening to just sections of a song and adjusting. Then you can of course burn disc, save to NAS or whatever & enjoy!

 

[karlkastner]

Over my head, but impressive. How about you work with Dutton Vocalion, producing synthesized 4.0 or 5.1 SACDs? Would bands accept/license this (re)treatment of their music?

I would buy them. Back In Black is my first request!

 

[Sonik Wiz]

Over my head, but impressive. How about you work with Dutton Vocalion, producing synthesized 4.0 or 5.1 SACDs? Would bands accept/license this (re)treatment of their music?

I would buy them. Back In Black is my first request!

Hi Karl! Thanks for the interest and reply.

The intent of my note was to manipulate a two ch stereo source before playing over a conventional matrix type decoder. The process does not create a 4.0 or 5.1 recording. The way I do that would be here:

https://www.quadraphonicquad.com/forums/index.php?threads/s2s-from-qs-via-pc-ok-part-1.5248/

Also I would never take copy righted material, upmix & resale in regards to the SACD's mentioned. And from what I know all of Dutton Vocalion is already surround sound & no upmix will (usually) be better than a professional release.

Although I use and prefer Adobe Audition there are many other software apps that this can be done with, some of them free. However Audition is the only one I know of with the dynamic Pan/Expand & Center Channel Extractor that comes in very handy.

 

[J. PUPSTER]

More fully: manipulating in 2 ch stereo on a PC to produce a desirable outcome when playing through a matrix decoder such as QS Variomatrix, etc.

For the most part when decoding stereo into surround I have had two ways of doing it. One is most casual such as simply playing through my Sansui QSD-1000, Fosgate Tate 101A, or DPL II on my Anthem AVM30. Otherwise I use the upmixing method I described several years ago on this forum for the really important music I wanted to enjoy in the best possible way. The latter gives excellent results but is very labor intensive. For oh maybe the last six months I have been sampling an intermediate method which is adjusting phase balance , EQ, panning, etc. on Adobe Audition prior to playing back on a decoder. It’s common knowledge that stereo into surround is quite variable & mostly it relies on how close the two channel source comes to matching the requirements of the decoder matrix. Controlling it this way puts a lot of power into making the original source do that. Even in the age of the Surround Master this is still valid since the SM does not have a synth or hall function as on old VarioMatrix decoders that can be quite useful. This method also works great when using SpecWeb at default settings. Also EQ can be applied. Doing out of phase blending can reduce the bass & that can be compensated for. I tend to listen to a lot of J Pop/Rock that is often pretty bright & some graphic EQ takes that edge off better than simply adjusting bass/treble controls.

The most useful tools in Audition for this is Channel Mixer and Pan/Expand (process) under Stereo Imagery. Channel mix can be used to create in phase or opposite phase mixing. Set on Wide Field it produces close to a -10dB opposite phase blend that idens the image in stereo & goes a long way to opening up the sound field when decoded. It is easy to make a preset that matches Sansui’s synth mode:
View attachment 34486
In this case a left only guitar in stereo will play back in left back with sharp directionality in QS , DPL II when adjusted right and I’m sure the SM too. A guitar panned left to right will make that beautiful 270 deg horseshoe enhancement that always adds so much to pop & rock.

The other tool I use is the Pan/Expand (process). I have never seen any comments on this tool mentioned before on the QQ forum so I wonder if anyone knows much about it. It widens the stereo field and provides out of phase blending but works much different than the channel mixer. On one hand it does dynamically make the dominant channel in the stereo mix even louder. It also creates out of phase blending towards the sub-dominant channel. Because it is dynamic it has most expansion when a single channel is present and zero action when the levels match as in a mono mix. This means you get opposite phase enhancement without a deficit of center front vocals or bass as you do when using channel mixer. A preset of 200 matches almost exactly the channel mixer Wide Field at -10dB blend and set at 250 you get the -7dB blend:
View attachment 34487
Neither is perfect I find one of the two might work better for a particular song or album. Pan/Expand will increase the signal levels so I usually Normalize at -6dB before using and eventually when all is done it’s my habit to normalize each song to -1dB. Maybe I state the obvious but you can also use these tools to mix in phase & bring the sound more up front. With the Pan/Expand you can create variable blending over selected time frame. That is you could choose a song that has a subtle mellow opening adjusting for full mono blend and gradually open up to original stereo & expand to a synth setting. This worked out very nice for G n’ R’s November Rain.

Also in Pan/Expand you could take (as an example) a solo lead guitar that is just in left channel & bounce it left/right in pan. Doing that and adding out of phase blend you now have a guitar that will bounce between left and right back or if slower it will pan around 270 deg.:
View attachment 34488
As mentioned before this is also a good time to do EQ. I like to tweak thing so my bass/treble can be set at zero. I’m not OCD about that; it’s just that it’s so easy to get more accurate tone control beforehand.

If this seems like a lot of trouble I can only say it’s much easier than my full blown method that takes hours to do on an album. My Delta 1010 (not LT) SPDIF goes straight to my Anthem AVM30. I can sit in the sweet spot; look at Audition on an eight foot screen and with a wireless mouse & keyboard I can hear exactly how it’s going to play back. I think it is pretty easy to make quick decisions listening to just sections of a song and adjusting. Then you can of course burn disc, save to NAS or whatever & enjoy!

Hey Scott, wanted to learn a little more on your pre-synth techniques AA3 (your screen shots here are low res.) so for the Pan/Expand preset do you just select a portion on both channels to adjust, then move the Stereo Expand slider up to 200 and create/name the Preset "200"?
I've added a better photo of where I was at, to make sure the settings look correct. And do you typically use 200 as your preset?

 

[Sonik Wiz]

Hey Scott, wanted to learn a little more on your pre-synth techniques AA3 (your screen shots here are low res.) so for the Pan/Expand preset do you just select a portion on both channels to adjust, then move the Stereo Expand slider up to 200 and create/name the Preset "200"?
I've added a better photo of where I was at, to make sure the settings look correct. And do you typically use 200 as your preset?

View attachment 43237

Yup you've got it exactly right. Label that preset 200 & save.
Technically the 250 setting gives the -7dB setting suggested by Sansui in articles, patents, etc. The 200 gives less out phase blend at -10dB. Your ears will tell you which is best for the song. Having said that I usually go with 200 because there is usually (song dependent) amounts of out of phase blend going on in stereo anyway.

I suggest normalizing a track to -6dB first because the Expand will definitely raise the level of the dominant sound, moment to moment. Otherwise it is easy to overload & clip.

Note that on Expand the bottom scale is 0 secs to X time. Whatever length you have selected, a portion of a song or all of it. Sudden jumps are not good within a song but you can also drag that curve around & make smooth transitions. For example if a song has a 20 sec intro, that gives you some time to start out at mono & expand to full synth surround in that time frame.

 

[par4ken]

Scott, I just recently came across this thread. The thought of preprocessing stereo this way had crossed in the past my mind but I never really gave it that much thought. It's quite a bit of work to pre-process everything and the QSD-1 and Photolune decoders already had the surround mode. As you know I also purchased an Involve evaluation board and have commented that it could use a surround and possibly a hall function. Your original Pre-synth circuit works fine for the purpose and the addition of the capacitor in series with the out of phase blend prevents the bass cancellation of the straight blending. That being said it still makes sense to process digital signals digitally as much as possible, so I've been playing around with this idea a bit. The Pan/Expand (process) seems very promising (if not a little weird). On the one hand it does not reduce the level of common (mono) information but on the other hand it increases the level of the dominant stereo signal (almost the same result?). One interesting thing I notice if you take a track from a typical brick walled CD such as the track in J. Pupster's example, after processing with the Pan/Expand (process) the result is much more dynamic looking the flatness of the original waveform is gone. While I don't think that the process can undo all the damage done to the track in the original processing it does seem to restore some of the lost dynamics.
Elsewhere I commented on the Steve Wilson mix of Roxy Music, while very immersive it lacks the punch of the original as the sax and synthesizer are buried in the mix. I put on the CD included in the package playing it through the QSD-1 in surround mode and while the surround effect was excellent, sound quality was terrible. I ripped the CD to see and it's brick walled and even clipped, TERRIBLE. By contrast my original CD sounds fine and shows no clipping or brick walling. I've run it through with processing at 200 and at 250 and will have to do more listening to see what setting I like best. While it would be a lot of work to pre process everything in this way it might be worthwhile in the end. Straight QS decoding of stereo leaves me flat, those that rave about the Surround Master should give this a try, I'm sure that they would be even more blown away.

 

[J. PUPSTER]

Scott, I just recently came across this thread. The thought of preprocessing stereo this way had crossed in the past my mind but I never really gave it that much thought. It's quite a bit of work to pre-process everything and the QSD-1 and Photolune decoders already had the surround mode. As you know I also purchased an Involve evaluation board and have commented that it could use a surround and possibly a hall function. Your original Pre-synth circuit works fine for the purpose and the addition of the capacitor in series with the out of phase blend prevents the bass cancellation of the straight blending. That being said it still makes sense to process digital signals digitally as much as possible, so I've been playing around with this idea a bit. The Pan/Expand (process) seems very promising (if not a little weird). On the one hand it does not reduce the level of common (mono) information but on the other hand it increases the level of the dominant stereo signal (almost the same result?). One interesting thing I notice if you take a track from a typical brick walled CD such as the track in J. Pupster's example, after processing with the Pan/Expand (process) the result is much more dynamic looking the flatness of the original waveform is gone. While I don't think that the process can undo all the damage done to the track in the original processing it does seem to restore some of the lost dynamics.
Elsewhere I commented on the Steve Wilson mix of Roxy Music, while very immersive it lacks the punch of the original as the sax and synthesizer are buried in the mix. I put on the CD included in the package playing it through the QSD-1 in surround mode and while the surround effect was excellent, sound quality was terrible. I ripped the CD to see and it's brick walled and even clipped, TERRIBLE. By contrast my original CD sounds fine and shows no clipping or brick walling. I've run it through with processing at 200 and at 250 and will have to do more listening to see what setting I like best. While it would be a lot of work to pre process everything in this way it might be worthwhile in the end. Straight QS decoding of stereo leaves me flat, those that rave about the Surround Master should give this a try, I'm sure that they would be even more blown away.

Hi @par4ken, I was also wondering about the dynamics after processing as you mentioned.
Here's 3 DR readings from the different progressions of the song "Groove Is King":

The first one is direct off the CD at 16/44.1

The next one is after Sonik Wiz's Pre-Snyth processing at 24/48:

And finally after I converted it from an analog feed through the Surround Master v2- Involve 4.0 QS mode- then through a MOTU UltraLite mk4 at 24/96:

So, what's going on here???

 

[J. PUPSTER]

Just some additional thoughts concerning the album Grooove is King by Rock Candy Funk Party.
Both @Sonik Wiz and myself noticed that this album (CD) decodes great in the Involve QS mode on the Surround Masters. I've always thought that the album has a very punchy mid to low end range to it. It also seems to have (from the looks of the wave forms) a brick walled look to it. But I never really noticed it being brick walled harsh that much just listening to it in it's original stereo form. I've also thought that the bass sound is very reminiscent of another bass player- Tony Levin (of Stick Men, Peter Gabriel and others.) The bass on Grooove is King is played by "Mike Merritt."
There is also an instrumental cover of the song "Digging in the Dirt" from the Peter Gabriel album "Us" on the Grooove is King album- coincidence?

But back to the overall sound of the album. With that punchy mid to low end presence and showing a more brick walled sound to it, you wouldn't think it would have the right phase material to decode well with the Involve QS. Perhaps by using the Pan/Expand process as @par4ken said "On the one hand it does not reduce the level of common (mono) information but on the other hand it increases the level of the dominant stereo signal", and in combination with the way the Surround Master decodes, to seemingly reduce some of that mid & low end punchiness results in a higher DR number.

Just for fun, here's a little vid. about the making of the album:

 

[par4ken]

Your DR results show that the processed signal has greater dynamic range than the original, which is what I noticed by looking at the waveforms. It's Interesting that the Surround Master seems to increase dynamic range as well, must be the result of adding and subtracting signals and whatever else the decoder does to enhance directionality. The brick walling is just compression to keep all the peaks all the same level, it should not effect phase, but it does impart a harshness to the overall sound quality. I find that I often have to turn down the treble on CD's, while I turn it up on analog sources.

 

[Sonik Wiz]

Well, yes everything you said is spot on. As experimentive as QQ'ers are I'm surprised that only you & the Pupster are working with this. And most happy to hear your feedback.

Scott, I just recently came across this thread. The thought of preprocessing stereo this way had crossed in the past my mind but I never really gave it that much thought. It's quite a bit of work to pre-process everything and the QSD-1 and Photolume decoders already had the surround mode.

IMO the last great pieces of quad gear for processing music was the Fosgate Tate 101A and the Sansui QSD-1000. I have read many good things about the Lexicon, which I have never heard, but out of respect I would give it the benefit of the doubt. But in general as the dark ages of Dobly matrix surround reared it's head and all audio gear became home theater stuff the Tate & Sansui's offered amazing performance. They both offered a synthesis mode for playing stereo and based on highly different approaches they both sounded great, they both sounded quite different. As my legacy gear failed and more stuff was going digital I pegged my hopes on DPL II which never gave me the same thrill. So in that magic age of Pentium 4 & Win XP I built my 1st PC devoted to audio & video. My motivation was to find a process that could duplicate Sansui's stereo to surround synthesis.

I definitely did this & in 2005 I posted a 3 part note on the forum how to do this. I don't know what happened too part 1 but part 2 & 3 can be found here:

https://www.quadraphonicquad.com/fo...roduction-and-playback-on-the-desk.53/page-16
Careful pre-synthesis was the 1st of 3 parts to the process, the 2nd being QS (more properly RM) decoding & any post processing as needed was the 3rd. This was very labor intensive & I used it for only the most important stuff including music video collections & a couple of feature length movies available only in stereo. I wanted a middle ground that would pre-synthesize for playing over DPL II & now of course the SM. It works quite well. The only drawback to the Sansui/Fosgate way of doing things is the synth is fixed. Doing it on the PC lets you vary the amount, song to song or even with in a song. It can be done super fast with scripts & batch processing. I suggest:
Normalize original at -6dB
Expand to 200 or 250
Normalize to -1dB

If you playback from the PC you don't even need to save or burn any further. My PC SPDIF goes straight tyo my Anthem.

One interesting thing I notice if you take a track from a typical brick walled CD such as the track in J. Pupster's example, after processing with the Pan/Expand (process) the result is much more dynamic looking the flatness of the original waveform is gone. While I don't think that the process can undo all the damage done to the track in the original processing it does seem to restore some of the lost dynamics.

Exactly & I'm glad you noticed this. You can both see & hear the difference. The Expand Process uses both in phase & out of phase blending that helps compressed audio as a side effect.

Straight QS decoding of stereo leaves me flat, those that rave about the Surround Master should give this a try, I'm sure that they would be even more blown away.

Chucky is of the opinion that there "is enough psycho-acoustic goodness baked into the Surround Master" that it doesn't need such stuff. On one hand I exactly see his point of view. The Sansui basic QS Vario-Matric & the SM operate very simple: any direction left/right is determined by the amplitude difference between the chs. Where it is located front/back depends on the phase difference between the chs. I guess you really can call it a purist approach. But on my other hand, the right one, I think there is more freedom to enhance stereo beyond conventional decoding. The SM gives a very wide presentaion & sometimes pin point sounds do poip up in back left/right. But having pre-synth really opens up the sound field. I am most happy with my little Chase RLC-1 that lets me fine tune level balance & phase balance precisley from the sweet spot. Except fpr you I don't expect many will get out the soldering gun to build any kind of hardware pre-synth but many will be able to experiment on the PC.

 

[Sonik Wiz]

Hi @par4ken, I was also wondering about the dynamics after processing as you mentioned.
Here's 3 DR readings from the different progressions of the song "Groove Is King":

The first one is direct off the CD at 16/44.1

View attachment 43278

The next one is after Sonik Wiz's Pre-Snyth processing at 24/48:

View attachment 43279

And finally after I converted it from an analog feed through the Surround Master v2- Involve 4.0 QS mode- then through a MOTU UltraLite mk4 at 24/96:

View attachment 43280

So, what's going on here???

That's really great data, thanks. I was wondering how folks here were coming up with dynamic range numbers such as 8, or 10, whatever. I see it is being done on Foobar 2000. I will look into this asa useful tool. Right now I use the amplitude window (goofy name) in AA 3 that gives me full data but not a simple DR number. Cool.

The stats are no mystery really. When the Expand Process works it dynamically increases the separation moment to moment between chs which also expands the transient dynamic range. Don't forget that Vario-Matrix techniques as used in the SM also does what? Increase the seperation between chs over what a simple matrix decoder would provide. It looks good to see the dynamic range is close to double in most selections.

 

[Sonik Wiz]

[

Just for fun, here's a little vid. about the making of the album:

Just had a chance to watch the vid. Good but would have liked to see more. Any other documentary stuff on the band? What's he say at the end? Presented in the Highest Infidelity? Love it!

 

[J. PUPSTER]

[

Just had a chance to watch the vid. Good but would have liked to see more. Any other documentary stuff on the band? What's he say at the end? Presented in the Highest Infidelity? Love it!

I'd have to check the DVDs that comes as a set with the CDs (also with- We Want Grooove)
Another fun little side trip - funny how what's being played doesn't match up with the video :

 

[Philip Spinner]

I'd have to check the DVDs that comes as a set with the CDs (also with- We Want Grooove)
Another fun little side trip - funny how what's being played doesn't match up with the video :

Ouch!

 

[ssully]

Sorry, what is this tool called? What is the upmixer you prefer these days for best results?

 

[J. PUPSTER]

Sorry, what is this tool called? What is the upmixer you prefer these days for best results?

I used to just use the Involve Surround Master in Involve (QS) mode; but because it's real time transfers, I made the switch to Penteo on my Win. 10 machine and get similar of better results and it's got more adjustment to the process. To process an average length song in Penteo only takes maybe 10-20 seconds. The pre-synth is just a sort of widener (IINM) to help stubborn songs throw more stuff to the rears before running it through an up-mixer. @Sonik Wiz knows more about this than I do. More recently though, I'm experimenting with iZotope Ozone Imager using Ultra Wide mastering or Stereoize for more widening, although sometimes it introduces a little more artifacts.

Also, in the last year I've been doing more De-mixing elements of the songs into isolated stems (De-Mix Pro)- i.e. vocals, guitars, piano, drums, other, etc..

There are just so many ways to go with this stuff. Always fun, to hear what other folks are using for their process.