PC Based SQ DECODING - ALMOST DONE!

QuadraphonicQuad

Help Support QuadraphonicQuad:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
QUADradial said:
You aren't wrong...but for some reason this approach works better. There are no pumping artifacts at all. Of course, none of this is in real time. From what little I've decoded, I prefer this approach over using a Tate.

The "dynamic" processing on the tate compensate for the lackness of logic discrimination - it's quite remarkable that using a Sq-matrix which in its origin had few dB of separation it was able to achieve a greater one. Unfortunately i think exactly that is the source of the "pumping" of any hardware SQ decoder. It was the only possible way to emulate a full phase decoding with the hardware that was available at the time.
This is also the reason why this approach works better: because amplitude is not a reference, just phase variations. Since SQ encoding was based only on phase, a pure-phase approach is more faithful to the original signal. Pumping and phasiness artifacts are possible even on PC, just set the amplitude on a value different from 0 and you can have both in the quantities you like.
In other words, there's no need of going the gain-riding logic and so on because it's possible to work *only* on the phase realtionships in a precise way. Math isn't easy with just electronics.

I'm not saying the Tate is now a piece of crap! The tate works in realtime, which it's still a bit far from this approach to be. Decoding nearly 1-hour of SQ (Lotus disc 1) took 8 hours on my machine (athlon 1800, kinda old nowadays...); but, once decoded, it's ready for a different encoding on actual digital media, and with a result which is better that anything else available until now.

The script i have posted works wonderfully on the Sq material i've ran until now. Tonight i will try a couple of Santana SQ (Caravanserai/Welcome) to check out how they compare against a TateSQ decoded *and* a 8 track discrete cart. They should be ready for tomorrow lunchtime...
 
Malcolm2010 said:
Neil these are the commands to load the relevant files, I do not have AA here, but one home I will check it out.

Malcolm

Malcolm.
Didn't explain myself properly again, methinks.
It is line 2 I have the trouble with - no way to type in what it says.
That sy(with the dots over it) and the square.
How do I enter that??
 
neil wilkes said:
Malcolm.
Didn't explain myself properly again, methinks.
It is line 2 I have the trouble with - no way to type in what it says.
That sy(with the dots over it) and the square.
How do I enter that??

Neil,

Just copy the entire script from here into Notepad, and
-> save it as "SQdecoding.scp" (w/o the .txt)
-> into C:\Program Files\Adobe\Audition 1.5\Scripts

I've done it that way and it works like a charm !

Regards,
Andreas
 
Thanks Andreas.
It is not going to be much use as yet due to the lack of ASIO support in AA - it will not recognize my HDSP9652 as a multichannel device.
But - I have certain hopes that this issue will be resolved very soon, so it is definitely going to be worth a look.
It's already a text file awaiting the happy day.
 
I couldn't help but notice one of the lines in the script referred to a value of SR set to 44100. I'm assuming this is the sampling rate. Is this so? Because I'm using 96kHz wavs, so I assume I'd have to change that value in the script.
 
kempfand said:
If I read this correctly, this is what the quoted operation does:
Lb = Lt - j*Ft
Rb = Rt - j*Lt
(with "j"= 90 deg phase shift). I think this is slightly incorrect, and does not correctly decode the non-logic rears for SQ feeds.

The specs say (see i.e. the Wendy Carlos link given above):
Lb = j*Lt - Rt
Rb = Lt - j*Rt
This excludes the x 0.707 / -3dB gain adjustments.

So in order to extract the rear information in Audion, I use:
1) Open the File in edit-view
2) Edit-menu < Copy to New
3) Edit Left Channel only ; Effects < Filters < Graphic Phase Shifter -> "+ 90 deg"
4) Edit Right Channel only ; Effects < Filters < Graphic Phase Shifter -> "- 90 deg"
5) Edit both channels ; Select Entire Wave ; Copy
6) Switch back to the original file opened under (1)
7) Effects < Amplitude < Channel Mixer -> "Swap Channel" and select "Invert" for "New Left Channel"
8) Mix Paste " (with "overlap") ;The result of this is the decoded non-logic Rears for SQ.

I've verified this with a channel-assignment testfile I created and encoded to SQ, and it de-codes correctly, whereas the the initial process outlines switches Lb vs Rb (and does wrong and probably other things as well).

Finally: In oder to correctly gain the rears, use Effects < Amplitude < Amplify -> 3dB cut just after step (1).

Andreas

Hi everyone out there! Stop decoding! Read this!

Of course I was wrong with the basic decoding matrix, Rb came out -90deg wrong, Lb came out +90deg wrong.
Instead of shifting the phase again, I looked closer to Wendy Carlos' vector calculations (see above posts of the hand drawn scheme and my handmade scheme below...)

This means, that two new files are needed to be mixed together:

File1: Rt +90phase, Lt -90phase, -3db reduced
File2: Lt original , Rt -180phase, -3db reduced,then channel swapped

If this two files are mixed together the real back "nologic" channels come out!

This means for the "center channel treatment": Remove the front center from the back center by removing at 180 degrees, not at 0 degrees!
I so and so doubted, why in the front it had to be removed at 180 phase, in back it was 0 phase with my wrong matrix. Now it is clear!

That there is no confusion with Andreas, please call me then Andy further on.

The resulting vectors are the same than at Wendy's drawings, with Lr and Rr accurate in phase!
I am sure that there are many different ways to get the same result, e.g. working with the old approach and then just change back the phases of the rear channels. I would suggest that for already done work!

Andy
 

Attachments

  • BasicSQDecodingV2.jpg
    BasicSQDecodingV2.jpg
    49.8 KB
Last edited:
ArmyOfQuad said:
I couldn't help but notice one of the lines in the script referred to a value of SR set to 44100. I'm assuming this is the sampling rate. Is this so? Because I'm using 96kHz wavs, so I assume I'd have to change that value in the script.

Sure! These are the values i've used for decoding some Sq-Encoded CD such as Lotus.
 
winopener said:
I'm not saying the Tate is now a piece of crap! The tate works in realtime, which it's still a bit far from this approach to be. Decoding nearly 1-hour of SQ (Lotus disc 1) took 8 hours on my machine (athlon 1800, kinda old nowadays...); but, once decoded, it's ready for a different encoding on actual digital media, and with a result which is better that anything else available until now.

I think realtime is possible...but with a programmed modern DSP processor. These are inside every surround amplifier that can do decoding of analog Dolby Surround or DSPII. They only have to be programmed to SQ phase filtering, not to DS. But who has a Zoran developer kit home and also the knowledge to program such device??? That is the thing. Would be too nice, just dreaming...
 
imploder said:
Hi everyone out there! Stop decoding! Read this!

Of course I was wrong with the basic decoding matrix, Rb came out -90deg wrong, Lb came out +90deg wrong.
...
If this two files are mixed together the real back "nologic" channels come out!

I've done a quick-and-dirty test with the sq test file.
Shifting the phase of the rear channel file done with the old methol of
left back: -90
right back: +90
and lowering the level by 3dB results in a file that can be inverted and sum to zero to the rear file done with the new method.

This means that, even with a 90 degree shift (which is easier and faster to correct than redo all the maths), there is no difference in the logic decoding. The separation that you get is the same, just a phase difference corrected.
 
winopener said:
I've done a quick-and-dirty test with the sq test file.
Shifting the phase of the rear channel file done with the old methol of
left back: -90
right back: +90
and lowering the level by 3dB results in a file that can be inverted and sum to zero to the rear file done with the new method.

This means that, even with a 90 degree shift (which is easier and faster to correct than redo all the maths), there is no difference in the logic decoding. The separation that you get is the same, just a phase difference corrected.

Tested the same with DSOTM "Money". Now the back channels sound more compact. I think, now it is really done!
Working on the vectors of QS...there are a lot...
 
Maybe all four final signals should be put in-phase for best listening results. The bass response should change.

Maybe all original signals from the needle drop should be phase-shifted to approach the original source.
 
neil wilkes said:
Thanks Andreas.
It is not going to be much use as yet due to the lack of ASIO support in AA - it will not recognize my HDSP9652 as a multichannel device.
But - I have certain hopes that this issue will be resolved very soon, so it is definitely going to be worth a look.
It's already a text file awaiting the happy day.

Hi Neil,

I use just the onboard soundcard (for test playback, not recording), that one is also not recognized als multichannel device in AA1.5.
I use Steinberg Nuendo for multichannel playback, that is ASIO compatible and with the ASIO4ALL driver it works also with my cheap onboard card.
This means, I hear the final result just at the end of the process...

Regards

Andy
 
Hi,
I have used this technique to decode SQ, and I must say it works well. I have a question for those who are experimenting, the major feeds front left/right, rear left/right front centre/rear centre decode perfectly. I have a problem decoding centre left and centre right. Using the first track from Quadrafile as a test, the centre left and right decoding leaves a substantial amount of information in the opposite channel. For example, when decoding centre left, at the end of the process only front left and rear left should remain. I find that a substantial amount remains in the front right and rear right channels as well, only about 9dB down from what is in the left. I hope this is clear, and I will try and post images to show what I mean. Further experimentation will happen tonight and I will see what can be done to try and remove this.
It is not a great problem, it would narrow the soundstage across the middle of the room creating a figure of 8 pattern, but would be nice to make it work.

Malcolm
 
Malcolm2010 said:
Hi,
I have a problem decoding centre left and centre right. Using the first track from Quadrafile as a test, the centre left and right decoding leaves a substantial amount of information in the opposite channel. ...

Malcom,
there's a way to get a quality-wise SQ test track? Since SQ is found 99.9% in LP, unless someone here has both a Quadrafile *and* a Laser player such as ELP, i suspect we would not find a pratical hint due to the lack of separation of the media. The CL/CR images are just 9dB, but there are a lot of leftovers on the 30-35dB area that if removed sound will be better for sure.
The only alternative will be passing a suitable 4-channel file created today from a PC to a SQ Encoder and sample it directly from the Encoder. That will give the best real-world SQ-encoded file for the tests - since real SQ LP were encoded with these hardware encoders and not with audio software.
 
winopener said:
Malcom,
there's a way to get a quality-wise SQ test track? Since SQ is found 99.9% in LP, unless someone here has both a Quadrafile *and* a Laser player such as ELP, i suspect we would not find a pratical hint due to the lack of separation of the media. The CL/CR images are just 9dB, but there are a lot of leftovers on the 30-35dB area that if removed sound will be better for sure.
The only alternative will be passing a suitable 4-channel file created today from a PC to a SQ Encoder and sample it directly from the Encoder. That will give the best real-world SQ-encoded file for the tests - since real SQ LP were encoded with these hardware encoders and not with audio software.

I also think so, the SQ test file I got from Tab at the very beginning (I think it was from a SQ tape?) has slight phase shift between the sum channels due to an incorrect adjusted playback head of the tape machine, also this file does not separate completely.
All files I digitized myself with an high end record player separate very, very well!
SQ was never intended to be separated again completely due to an AES paper from CBS I have in hands, because also the phase shifting while encoding was made with analog equipment and not all frequencies are shifted exactly the same angle.

As I already wrote in this forum: The better the encoding was and the better the playback hardware is, the better the result will be.

One important thing: If r and l channels are swapped accidently while recording, the extraction of the rear channels will partly fail ! Looking at the phase vectors it gets clear.

Next topic: Why it could not have happened that also some encoding processes were made wrong, especially while the early "non-logic" area of SQ as the errors could not be heard?
 
imploder said:
SQ was never intended to be separated again completely due to an AES paper from CBS I have in hands, because also the phase shifting while encoding was made with analog equipment and not all frequencies are shifted exactly the same angle.

that's why if you use the strongest settings possible all that you have back is crappy phased sound; widening a bit the tolerance allow for a decent shift of the main 90 phase angle without trashing the recording and the separation.

imploder said:
Next topic: Why it could not have happened that also some encoding processes were made wrong, especially while the early "non-logic" area of SQ as the errors could not be heard?

I think it had happened... IIRC some months ago surfaced some SQ copies pressed in Brasil of a ELO album, strangely released only there, and when decoded the front right channel was nearly absent.
 
If it was "Chase," then it was from a factory CD. If it was the quadraphile SQ test tones, then it was Cai's CD made on a very precise turntable.
 
I may do the jump to a good audio card for needle drops and SQ decoding is something I'd love to try.

¿Can I use a good two-channel audio card for recording and an inexpensive Soundblaster Live 5.1 for analog playback from Adobe Audition? I use W2000.

Thanks in advance.
 
Interesting... Files decoded from SQ using the software method can easily be re-encoded into DPL-II with very good results. Files that were Tate decoded cannot.
Also, I am wondering if pushing the separation past the cartridge rating on vinyl played SQ is causing some artifacts. I think if you try to get more separation than the stereo cartridge could provide might be a mistake. My cartridge is rated at around 25db left/right. Pushing the software past that makes it sound watery during the most quiet passages...
 
Exactly! Besides the separation of phono cartriges goes down at higher frequencies (mine has >25db at 1KHz but "only" 15db at 15KHz, and it is a good MC cartirge. So if we go beyond the limits of the original, we get out "digital" reults.
Just decoded "Johnny Cash at San Quentin". Nice atmosphere!
And finally I will decode, what brought me to Quad: Ennoch Light...
 
Back
Top