Listening to this surround UPMIX, up-remix, or remix

QuadraphonicQuad

Help Support QuadraphonicQuad:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Wouldn´t it be interesting to set-up a competition/comparison of upmixing method's/equipment. We will have to find a piece of stereo music that is free of rights, and publish it here. And than everybody uploads their upmix here for comparison/judgement.
 
You may already be aware, but Pretzel Logic was released in QS. Oxford Dickie did a decode of this using his scripts some years ago. It would be interesting to compare your upmixing to his decode. What source were you upmixing from? If it was the matrix version rather than standard stereo then the results may vary. I use DTD neural for my upmixing, but haven't tried comparing results from stereo vs matrix sources.
The upmixes of Pretzel Logic we’re from a high resolution stereo source. The quad that is out there sucks...like double stereo only worse IMO.
 
The upmixes of Pretzel Logic we’re from a high resolution stereo source. The quad that is out there sucks...like double stereo only worse IMO.
Agreed RE the QS version of Pretzel Logic - It truly is awful. I was curious though whether there would be a difference in upmixing from a SQ or QS encoded source vs a 'standard' stereo source. To this end, I grabbed PBTHAL's vinyl rip of the QS matrix and ran it and a high res stereo source (SACD) through DTS Neural to see what the results would be like. It is like chalk and cheese... The QS version achieves virtually no separation (i.e. double stereo as you noted above), whereas the SACD source gives quite a good discreet upmix. I then thought it would be interesting to see what happens with some other source material - so I gave various versions of 'Money' from DSOTM a go through DTS Neural. For this I have available PBTHAL's rip of the SQ Vinyl, plus the 1973 and 2011 stereo versions from the Immersion box set and the original 4.0 and 5.1 mixes (also from Immersion). The results were really interesting - as the 40 second clips here show Pink Floyd - Money Source Comparison (starting from when the bass comes in).

The SQ mix through DTS Neural still has quite a bit of vocals in the rear starting at about 28 seconds, whilst the 1973 and 2011 have virtually none (1973 is slightly less than 2011). The kick drum is pretty much absent from the SQ upmix, and is stronger in the 2011 version than 1973. To see how all three compared to the discrete surround mixes (1973 4.0 and 2003 5.1) I have also included extracted rears from both. Of these, the 4.0 mix is very discreet, while the 5.1 is very busy compared to all the others. Using the ABX plugin for Foobar, you can switch between all five versions to hear the differences

What would also be good to compare would be a decode of the SQ Rip against the discreet 4.0 Immersion version, but I don't have the necessary scripts to do this.
 
Agreed RE the QS version of Pretzel Logic - It truly is awful. I was curious though whether there would be a difference in upmixing from a SQ or QS encoded source vs a 'standard' stereo source. To this end, I grabbed PBTHAL's vinyl rip of the QS matrix and ran it and a high res stereo source (SACD) through DTS Neural to see what the results would be like. It is like chalk and cheese... The QS version achieves virtually no separation (i.e. double stereo as you noted above), whereas the SACD source gives quite a good discreet upmix. I then thought it would be interesting to see what happens with some other source material - so I gave various versions of 'Money' from DSOTM a go through DTS Neural. For this I have available PBTHAL's rip of the SQ Vinyl, plus the 1973 and 2011 stereo versions from the Immersion box set and the original 4.0 and 5.1 mixes (also from Immersion). The results were really interesting - as the 40 second clips here show Pink Floyd - Money Source Comparison (starting from when the bass comes in).

The SQ mix through DTS Neural still has quite a bit of vocals in the rear starting at about 28 seconds, whilst the 1973 and 2011 have virtually none (1973 is slightly less than 2011). The kick drum is pretty much absent from the SQ upmix, and is stronger in the 2011 version than 1973. To see how all three compared to the discrete surround mixes (1973 4.0 and 2003 5.1) I have also included extracted rears from both. Of these, the 4.0 mix is very discreet, while the 5.1 is very busy compared to all the others. Using the ABX plugin for Foobar, you can switch between all five versions to hear the differences

What would also be good to compare would be a decode of the SQ Rip against the discreet 4.0 Immersion version, but I don't have the necessary scripts to do this.

I have dts Neural Upmix on my PC but I have been able to get it to work with Adobe Audition 3. I suspect it can't handle the newer version of VST that the dts uses. May I ask what you are using with dts Upmix?
Thanks!
 
I use the VST adaptor plugin for Foobar (foobar2000:Components/VST 2.4 adapter (foo_vst) - Hydrogenaudio Knowledgebase) to listen in real time and adjust settings, then I have some custom batch scripts I wrote years ago to process the files using a command line based VST host called Mrs Watson (MrsWatson). Once the files have been processed, the batch file splits the output into separate files, automatically checks/balances levels (by default it normalises the rear channels to -6dB compared the fronts) using Sox (SoX - Sound eXchange | HomePage) before recombining into 'final' flac files. It's just then a quick re-tag using Foobar and they're ready to enjoy. The whole process is pretty quick - about 15 mins per album.
 
Can we exchange the SpecWeb parameters how to come to the upmix result. Are you willing to share?

Verstuurd vanaf mijn SM-A500FU met Tapatalk
My use of SpecWeb is only using the basic tools available. I did try dabbling with other parameters and more fiddle factor but decided I was getting an acceptable result just by copying the wav flacs and pasting them straight into SpecWeb and then pressing Enter. That's it.

Maybe with all this time on my hands I will spend more time fiddling with the mixes but first I need to rip a ton of blu-rays (keep putting it off :rolleyes: )
 
My use of SpecWeb is only using the basic tools available. I did try dabbling with other parameters and more fiddle factor but decided I was getting an acceptable result just by copying the wav flacs and pasting them straight into SpecWeb and then pressing Enter. That's it.

Maybe with all this time on my hands I will spend more time fiddling with the mixes but first I need to rip a ton of blu-rays (keep putting it off :rolleyes: )
I do the same. Sounds a little thin compared to the stereo.

Verstuurd vanaf mijn SM-A500FU met Tapatalk
 
For those members who are not thinking about up-mixing or concerned about starting the process, may I take this opportunity to highly recommended the SpecWeb route. It is a free software package although there is a donation, should you wish to make one.
I was daunted by the perceived process and I am pleased to say that my initial concerns were misplaced as the actual method required is fairly straightforward to use.

I realise there are other up-mix packages available but for an amateur, like me, I am getting results way above my expectations, with SpecWeb

The important rule in all this work, is to seek out the best recording you have, to give it every chance of producing the result you are after.

It is very true to say what goes in determines what comes out.

To break the process down I have given a simplified list:

Load SpecWeb to your desktop.
Rip your required music into a flac format.
Use foobar to convert the flacs into wavs
Saving these into a new folder.
Load the wavs into foobar.
Copy them
Paste the required track/album into SpecWeb
Press enter
Wait for the process to finish
It will load the new mch flacs into the wav folder.
Sit back and enjoy your results.

There are plenty of ways to change the parameters in how you want your mix to sound but this is the way I keep coming back too.
 
For those just getting into upmixing, it's very important to state that you should try to find sources that are not brickwalled. Many times an earlier release CD will not be, while a "remastered" or late release CD of the same album is bricked to the max.

The upmixing software, be it SPEC, SPECWeb, Penteo, etc, works much better on a source that is not brickwalled.

Here's an example of the same song, the original CD and the newly remastered CD, and in this particular case, the brickwall is not that bad. Still, not having the loudness cranked helps the upmix.

Guess which is which?

brick.jpg
 
I don't use it much anymore but I will certainly agree SpecWeb is the eaisiest way there will ever be to upmix on a PC. And the quality can be very good. But the SM has reduced my need for any PC upmixing.

Load SpecWeb to your desktop.
Rip your required music into a flac format.
Use foobar to convert the flacs into wavs
Saving these into a new folder.
Load the wavs into foobar.
Copy them
Paste the required track/album into SpecWeb
Press enter
Wait for the process to finish
It will load the new mch flacs into the wav folder.
Sit back and enjoy your results.

It seems there more steps than needed. I would just drag a WAV file onto the SpecWeb desktop icon & let 'er rip. It comes back a nice 5.1 FLAC file that I would play through my Oppo.

I'd also like to point out SpecWeb has the fewest artifacts when using high bit rate & sample rates. I would upload 32 bit floating point at 96 & they came back as 24/96 file very clean.
 
I don't use it much anymore but I will certainly agree SpecWeb is the eaisiest way there will ever be to upmix on a PC. And the quality can be very good. But the SM has reduced my need for any PC upmixing.



It seems there more steps than needed. I would just drag a WAV file onto the SpecWeb desktop icon & let 'er rip. It comes back a nice 5.1 FLAC file that I would play through my Oppo.

I'd also like to point out SpecWeb has the fewest artifacts when using high bit rate & sample rates. I would upload 32 bit floating point at 96 & they came back as 24/96 file very clean.
Cheers Sonik
All hints n tips gratefully received.... (y)
 
Back
Top