S2S from QS via PC, OK? Part 3

QuadraphonicQuad

Help Support QuadraphonicQuad:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

Sonik Wiz

👂 500 MPH EARS 👂
QQ Supporter
Joined
May 30, 2005
Messages
5,304
Location
Kansas City
Having started somewhat in the middle
of this S2S process, with the decoding, it’s time to see what can be done with pre/post enhancements & variations of the decoding. I sperated it this way to make clear that the following methods are not integral to the decoding.

All righty then.

PRE ENHANCEMENT
A PM I received from Lucanu that had tried this decoding process as described so far remarked that it “did not sound very quaddy.” He is exactly right for two reasons:
1, I gave wrong values for Preset #3 in Channel Mixer
2, It is essential for pin point left back / right back sound to use a pre synthesis blend before decoding.

Using only the decoding as previously described does give some good surround effects, in fact for certain pieces it might be preferable to aggressive wrap around surround. It is very close to the sound of basic QS decoding with Vario Matrix, giving more of a center left & center right decode for left only or right only stereo inputs. Not bad on its own, nice to have a choice.

To make a left only stereo input decode as left back only, the decoder needs to “see” this as a left back signal. The QS PRE SYNTH preset made in Channel Mixer does this. For a left only signal the preset mixes the left audio to the right channel at 7.7 dB lower and polarity inverted compared to the original. For a right only the situation is reversed: the right channel is at full level and the left channel now carries the signal at inverted polarity, 7.7 dB lower.

This is the same approach done on all Sansui Vario Matrix decoders, but they were able to do it with only a simple resistor switched in between two transistors to accomplish this in hardware. In this PC adaption, using the pre synth blend will produce a rock solid image in left back or right back speakers. Sounds dynamically panned from one channel to the other in the stereo original will start in a back channel and move in a semi circle around you & end up in the other back speaker. Center front stays center front. Natural ambient reverb stays behind you. It is this capability of my S2S approach that sets it apart from many other methods that are available on the PC as plug ins or other methods talked about on the Net.

Obviously, you don’t have to use this blend at all or may choose to modify it for your own needs. After several listening tests I have settled actually on using a .3, or -10dB, opposite phase blend for my own projects. This is listed as simply “Wide Stereo” in the AA Channel Mixer preset library. I have made a demo DVD with both music video clips and music only tracks & have sent it to a few members of the QQ Forum. On all except two tracks I have used the Wide Stereo .3 blend. So for those that have heard it, consider that the left back / right back action could still be kicked up a notch. I use the lesser blend for two reasons:
1, Any out of phase blending to the stereo original does take a toll on the common mode audio, i.e., usually vocals and bass, and attenuates them. So the least possible blending of this type to achieve a given effect the better.
2, A .3 blend takes into account the amount of natural out of phase content that already exists in the stereo original and still locks on to that 270deg wrap around sound quite nicely.
The higher .4 blend, as in the QS Pre Synth preset, sounds great on high separation instrumentals where there are no vocals to worry about and reduction in bass can be compensated for by EQ or raising the LFE level. Some Herbie Hancock tracks are on my short list to do this to.

Think of the creative fun you can have with this pre synth. You can choose different amounts for different songs, or choose varying amounts within a song. You can also apply an overall pre synth blend to a song, select the intro portion of the music, then go to AA Pan Expand effect. Choose narrow (zero separation) at the beginning of the selection’s time and Expand to normal, 100, at the end. The song will start out on mono center front and expand in width & depth to full surround at the end of the selection. Using the Graphic Phase Shifter you can have a sound pan from front to rear. Start by making a selection you want to pan & choose left only. Go to the GPS and select 0 deg phase change at the beginning and +90 deg at the end. Then select right only and do the same but have it go from 0 deg to -90 at the end. Effects like this are easy to do, not so easy is make smooth transitions between the effected part and the rest of the music. Choose your edit points wisely. And sometimes it is better to have a pan, phase shift, or expansion effect start at zero or neutral, advance to the intended effect quickly, and then return to zero at the end of that time frame.

POST ENHANCEMENT
Only once have I ever heard the remark made that listening in the stereo surround modes on a Fosgate 101 or Vario Matrix decoder seemed to diminish & flatten the front soundstage. Maybe we are all just too fixated on the back channels to catch what’s happening up front sometimes! But I agree totally with who ever made that observation. It’s understandable that if you take the hard panned left & right signals from a stereo mix & put them behind you, then they simply are not there any longer to aid in the width of the front soundstage. In theory a certain phase/amplitude combination is still going to equal decoded left front or right front only, but in practice it just doesn’t work out that well. The fix to this is some simple post processing of the front channels.

After pre synth, decoding, and exporting, put the left front wav and right front wav together as a left/right stereo pair in the AA Edit window. Remember that unless you have renamed the exported files you will be using left front export & right surround export. When both are together as a stereo pair just go to AA Pan Expand effects and choose Expand. The Expand function is very useful and works by giving increasingly more amplification to the audio as the signal becomes more progressively left or right. At first glance it seems to have some attributes of CCE but is actually quite different. Expand does not remove or reduce any information as CCE does. It simply accentuates the level changes between two channels. I find that using a .3 Wide Stereo pre synth, decoding as described, that usually a 200 setting on Expand returns the front stage width on par to what’s occurring in the rear signals. When you do something like this be VERY CAREFUL how you save your files when you’re done. Select the entire LEFT ONLY portion for left front. Click on FILE and SAVE SELECTION (NOT just save or save as). Navigate to the where the exported left front file is saved and select that file as the name to be saved. Save as that name & location. AA will ask you twice if you want to write over the existing file. Say yes, and the exported left front file will be replaced with the separation enhanced version. Repeat the process with the right channel, saving that selection as whatever name and location the exported right front file has. This is also the basic way & point in the process you would make any individual level changes or EQ to a channel.

At some point it will hit you that even though you are not working with a true discrete multi channel mix, you still have four or more decoded files waiting for your creative touch. What do you want to do? I’ve given a few hints here but the combination of pre synth, decoding, and post manipulation is almost limitless.

VARIATIONS ON A THEME

Yes, you can even change the actual decoding parameters to suit your own needs & sonic goals. When I started experimenting with this PC S2S I chose the good ‘ol Sansui blend factors of .41, or 7.7 dB, out of habit. I also experimented with a lesser decoding blend of .31 and a higher one of .5. To compliment these changes I also tested matching pre synthesis blends. In theory any symmetrical matrix will respond similar manner for both pre synth and decoding crosstalk, and so the CCE would work equally well with symmetrical crosstalk. What happens to the center channels when applying diagonal crosstalk reduction is a different story. The .3 matrix gave a wider left/right feel with less front to back separation. The .5 matrix was just the opposite. The front to back sound fields were great but it always sound very closed in left and right, and the Expand effect couldn’t fix it with out adding it’s own problems.

So back I went to a .41 symmetrical decoding matrix as a model. You can still do some useful changes to it. For example when in Channel Mixer & your doing the QS FL/Rs & QS RF/Ls functions you can set the front channel blends to zero. This will give you pretty close to the Sansui mode of decoding rear channels only, with crosstalk reduction between them for a very nice & natural, non-wrap around effect. This worked great for me on a blues jazz stage performance where the front soundstage was just right but I needed the ambience, minus vocals, in the rear. Another variation is to decode but do not apply the CCE crosstalk reduction. Instead you can try the Expand function again to simply accentuate the difference between the LF/Rs and RF/Ls channels. And of course you do not have to apply the pre synth at all in some cases. A good example is a song I processed that was a pseudo live performance, much too much reverb added to make it sound like a BIG auditorium. Using a wrap around pre synth would have just made it worse. Doing just the regular decoding and the crosstalk reduction removed much of the excess echo from the front channels & put it where it belongs, behind the listener. This also resulted in a much sharper focus for the front soundstage.

FINAL THOUGHTS

To quote Peter Scheiber, “those who feel at home in multidimensional audio space are encouraged (dared?) to experiment”. Synthesizing stereo to surround on a PC offers a universe of potential for experimentation. Many of us have hung on to our old Quadraphonic decoders for the practical reason that there is not any other way to correctly play back music we like from a format we like. Also, even compared to Pro Logic II or comparable methods, the old hardware was still the best for performing the magic trick of creating surround sound from stereo. By my method for S2S and others looking at decoding Quad formats on a PC, the old boxes have reached a point where they have only nostalgic value or offer simply the convenience of instant playback. Decoding on a PC even eliminates the questionable value of multi band decoding since the PC does not operate in real time, it simply process’s on a sample to sample basis, artifact free. Not so many people are lucky to own the really good Quad decoders but so many others do own a computer! On a PC you can put your own individual touch on your music. You can easily use a 24/96 source, decode to surround and burn a DVD-A disc in lossless compression. I used to spend many hours with a soldering gun in my hand, building audio circuits to do what I wanted with music. Now I hold a mouse and do my building on my computer. It’s all worth it when I sit down it that sweet spot, & push the play button.

“May all your good dreams & fine wishes come true”, M. Jittlov ‘89
 
Amazing stuff, Sonic! You have hit the Holy Grail as far as I'm concerned. Kind of overwhwelming for me to try and actually do, though. I sure wish that you could release your work as commercial software. I'd buy it for sure.
 
To Lucanu & BrownRB:
Thanks guys, really! The silence was deafening there for a while. I have really enjoyed the various PM's from Lucanu & others. Hopefully a few are off & experimenting with their own music.

For those that own Tate or a QS decoder, I'd really like to hear how you think stereo decdoing on a PC compares.

I did "In Search of the Lost Chord" a coup[le of days ago. In the track "Thinking is the Best Way to travel" there is a medium pitch sound dynamically panning around with a higher pitch bird like sound dynamically panning at a different speed & completely different direction. The S2S allowed them both to fly around the room sharply focused & independently with out any artifacts. That would be tough for even a multi band decoder like the D1.
 
Well, I've figured it out (at least following your suggestions - not exploring on my own yet). This is a bit addictive. The first track I tried, The Dandy Warhols - Last High, wasn't doing quite as much as I wanted. So, I tried some of In Search of the Lost Chord - much more interesting. Now I'm off trying other tracks :) Thank you for posting all of this information, I would never have been able to figure this out on my own. (y)
 
BananaSlug said:
Well, I've figured it out (at least following your suggestions - not exploring on my own yet). This is a bit addictive. The first track I tried, The Dandy Warhols - Last High, wasn't doing quite as much as I wanted. So, I tried some of In Search of the Lost Chord - much more interesting. Now I'm off trying other tracks :) Thank you for posting all of this information, I would never have been able to figure this out on my own. (y)

Yes, it is addictive! I'm glad your efforts are turning into some fun. I don't know about The Dandy Warhols. I've heard of them. One thing to remeber when messing with S2S is that ultimately the basic quality of the stero recording will always be there. That means that sometimes it will never be exactly what we want. For example, music simply must have an amount of hard panned left & right sounds if you want something distinct to pop up in the rear channels. One thing to try is some gentle Expand on the Pan Expand effect to widen the left & right sound field. This does not effect or detract from center front vocals but it may throw the balance to the higher frequencies making it seem a little bright. Try this on the basic 2 ch source before any pre synth or decoding.

See Ya Bye!
R. Scott
 
Sonik Wiz said:
After pre synth, decoding, and exporting, put the left front wav and right front wav together as a left/right stereo pair in the AA Edit window.[/COLOR][/I]

O.K. I feel silly asking this - there must be an easy and probably obvious way - how do you put 2 mono .wavs in the edit window? I tried cutting and pasting to a stereo .wav (LF Rs), but the pasted in (RF) .wav came in at a higher volume?! - at least by looking at the graph.
 
BananaSlug said:
O.K. I feel silly asking this - there must be an easy and probably obvious way - how do you put 2 mono .wavs in the edit window? I tried cutting and pasting to a stereo .wav (LF Rs), but the pasted in (RF) .wav came in at a higher volume?! - at least by looking at the graph.




Well, if it’s that simple, I wouldn’t have been totally confused the first half dozen times I tried it too. I work pretty often with at least two apps of AA 1.5 open so here’s how I do it. Others might have a better way to suggest.

Open your first mono wav ( let’s say Left Front) in AA 2 ch edit window as you would normally do. Select all & copy.
Minimize this first app of AA.

Click on the AA shortcut/icon/whatever & open up a second app of AA. Some programs will simply restore the currently open app but AA will open up a legit second running program.

Click on FILE / NEW and select STEREO and the appropriate bit/sample rate.

At the middle of the top toolbar are three little icons to select:
Edit left only
Edit right only
Edit both channels

Select LEFT ONLY
Edit / Paste

Ta dah! Halfway there.

Go back to the first app of AA.
Click on FILE / CLOSE
Do not click on Close all or it will also close your copied wav.

Then open your second wav, say Right Front (which would be exported as Rs, unless you changed the name)

Copy.
Minimize the first app of AA again. Go to the second app that has your left wav in it.
Select the EDIT RIGHT ONLY button.
Edit / Paste

That’s it! Now you can go back to the first app & close out the Right Front wav that’s there, return to the second app & go at it.

Here’s a part I want to repeat to make the saving of these files hopefully confusion free. After whatever editing you need to do, go make sure you select LEFT ONLY first, then go to FILE / SAVE SELECTION. Navigate to where the Left Front exported file is and select that to save as. This will save the left front back as the wav file they started out as, but modified with your edits. Even if you select LEFT ONLY and FILE / SAVE or SAVE AS this will result in combining the individual wavs back into a stereo file. It must be SAVE SELECTION and be sure the entire length of the wav is selected.

Random thoughts on opening multiple apps of Adobe Audition

Scripts are a good way of speeding up repetitive & probably time consuming tasks in AA. It is also a good way to avoid making a mistake. And it’s a good way to NOT give your audio the attention it deserves. The more you batch and automate the less chance you have to fine tune your surround sound to exactly how you want it.

So I tend to pick up speed by opening two or more running programs on AA & switching back & forth while they work. A good example of how this helps is when the crosstalk reduction is done with CCE. On my computer it takes almost exactly 1/3 of the time of the wav it’s processing to complete. So a 60 minute wave will take 20 mins to process. I will open up two apps & get one going on CCE doing LF/Rs and the second one doing RF/Ls. Yes it does slow it down a little bit, but it only adds about 4 mins to the time. This is including click time to get the second part going. This means I have actually saved a net of 15 mins. You could also get CCE going to create a Center front channel, while you work at the same time on filtering the sub woof channel.

Results may vary when trying this. Some computers may lock up. For myself it works fine & I seem to be able to keep track of the steps better than if I just let a script do it for me.
 
Back
Top