SPECWEB (Now 2.2)

[zeerround]

no need to uninstall.

I just delete the icons on the desktop and then run the new installer.

You can ask about the helper in the helper thread.

 

[JanBakker]

no need to uninstall.

I just delete the icons on the desktop and then run the new installer.

You can ask about the helper in the helper thread.

Oké thanks, it works with helper it just needs the location of specweb.exe

 

[HomerJAU]

I will release Helper 1.3 later today (see Helper thread for updated info).

 

[zcftr29]

For the new version, I suggest you try the following command line options:

-Z -6-3

That will use "normonly gains" for ZAG (assumes your speakers are calibrated to all have the same volume level at the listening position) and use a Mastering Limiter, with a threshold of -3db, relative to the song's peak level.

The idea is to more closely match the loudness and levels of the original stereo.

If you still want to control the channel to channel levels yourself, you can either keep -Z and use the gain controls, or you can skip -Z and use the ZAG channel gain controls.

...

If your 5.1 speakers are not in the ITU configuration (as to angles) in this new version of SpecWeb you can tell it what your angles are (see the guide and included angle chart).

Cheers,
Z

Wow... more options to play with!!! I have been struggling for the last few days with getting levels sounding correct on Thriller (the album, not the song). 'Beat it' is proving particularly difficult. I'm getting some really good separation, but some of the elements of the mix a getting muted, while the vocals are coming out too strong. I'm still aiming to get quad upmixes rather than 5.1 - and I have had some really good results. Once I have got to grips with the new settings available I will report back with my revised process (which I hope will be a bit simpler)

Thanks again for all your hard work on this great bit of software...

 

[zeerround]

Wow... more options to play with!!! I have been struggling for the last few days with getting levels sounding correct on Thriller (the album, not the song). 'Beat it' is proving particularly difficult. I'm getting some really good separation, but some of the elements of the mix a getting muted, while the vocals are coming out too strong. I'm still aiming to get quad upmixes rather than 5.1 - and I have had some really good results. Once I have got to grips with the new settings available I will report back with my revised process (which I hope will be a bit simpler)

Thanks again for all your hard work on this great bit of software...

You are most welcome. Your input has inspired some of the features in 1.3. Namely the surround sound field rotation, and the addition of speaker angle settings. With those (and setting the center width to zero) I would hope you can do single step stereo to quad conversions, even for those odd songs with vocals panned hard left, etc.

Yeah, also try the postzag master limiter to help with your "muted elements" issues.

 

[Eye of Horus]

Something interesting I think. I'm sure those who "understand" will give me a reason...but this is striking.

So, I have a DSD64 download of Turn Of A Friendly Card - Alan Parsons. I'm sure any of you who have this know how amazing it sounds. It's as good as it gets....
So, to get it to work with Specweb, I made a duplicate copy of the DSD files and converted the duplicate to Flac...then I let Specweb do it's thing.

There really isn't much going on in the rears in terms of isolation....I assumed Specweb would have a hay day with this one....but no, it didn't. Yeah, it sounds great, but what is mostly in the rears is some echo/reverb from the information in the fronts. Again, I'm super surprised...

Now - does that have something to do with the fact that the "mother" files are DSD? I dunno...someone more educated in the music can comment.

Again...I found this very interesting.

The Turn of a Friendly Card is my alltime favorite album from my alltime favorite band !
I used the HDAD tracks for my conversion and my conversion shows a lot more in the rears than you mention here.
But... I used another method, WAMINU.
Together with the other HDAD's from the APP, I made a Bluray containing the 24/96 in stereo and in upmixed 5.1 DTS HD MAS 24/96.
It sounds great and the bluray gets playtime every week ;-)
Also got a lot of positive feedback on this bluray.
The other titles on it are I Robot and Eye in the Sky.................

One thing I want to mention, nothing to do with the conversion : it's The Alan Parsons Project, not Alan Parsons solo. That's quite a difference !
The Alan Parsons Project was mainly Eric Woolfson ! He wrote and composed all the songs and played with the musicians in the studio....
Alan was the producer. His name is only also in the "writers" credits for royalties sake ;-)
Eric made the project more than Alan and if you listen to Alan's solo albums you will without any doubt notice big differences !

grtz,

EoH

JFYI : My nickname came from the symbol of the Eye of Horus on the cover of Eye in the Sky ;-)

 

[zeerround]

Something interesting I think. I'm sure those who "understand" will give me a reason...but this is striking.

So, I have a DSD64 download of Turn Of A Friendly Card - Alan Parsons. I'm sure any of you who have this know how amazing it sounds. It's as good as it gets....
So, to get it to work with Specweb, I made a duplicate copy of the DSD files and converted the duplicate to Flac...then I let Specweb do it's thing.

There really isn't much going on in the rears in terms of isolation....I assumed Specweb would have a hay day with this one....but no, it didn't. Yeah, it sounds great, but what is mostly in the rears is some echo/reverb from the information in the fronts. Again, I'm super surprised...

Now - does that have something to do with the fact that the "mother" files are DSD? I dunno...someone more educated in the music can comment.

Again...I found this very interesting.

I grabbed the SHM CD version (I've always had good luck with those). Using SpecWeb Play, I worked with "Games People Play", and adjusted the widths to push more stuff to the rears and make sure things sounded good in each of C, Fronts, and Rears. I increased the rear blends a little to clean those up.

Try the attached ini file (with SpecWeb 1.3) with "Games People Play" and "Time" and see what you think. I did this quick and dirty but the idea is to give you a tip as to how to deal with tracks like these.

Edit: 3-27 Replaced attachment with the correct ini file.


Cheers,
Z

 

[Eye of Horus]

Downloaded the .ini file to see what came out.....
I took "Time" from the last remaster double CD.
My results with your settings surprised me..... unfortunately not in a positive way.
Could be your setting or something wrong on my side.
After the conversion I loaded the mch-flac in SF and measured the PEAK and RMS values, because the picture of the file already showed there was something not quite right.

The values :

LF/RF : RMS : -28.7 PEAK : -5.1
C : RMS : -17.8 PEAK : -0.2
LFE : RMS : -30 PEAK : -9.1
Ls/Rs : RMS : -34 PEAK : -11.1

This is what came out of the .ini and those values speak for themselves..........

After some work to get those values more in line, the artefacts in the rears became quite terrible.

Did I do something wrong or was there something wrong in the .ini file ?

 

[zeerround]

@EOH, did you use specweb 1.3 (released yesterday)?

 

[zcftr29]

You are most welcome. Your input has inspired some of the features in 1.3. Namely the surround sound field rotation, and the addition of speaker angle settings. With those (and setting the center width to zero) I would hope you can do single step stereo to quad conversions, even for those odd songs with vocals panned hard left, etc.

Yeah, also try the postzag master limiter to help with your "muted elements" issues.

I tried to use version 1.3 but it keeps on crashing... I started using the default 'play' icon - set values via the web interface and saved an 'ini' file. I made a couple of tweaks to the saved ini file (see attached), but when I then try to run the conversion, it crashes every time... I have tried this on two separate computers with the same result - both are running Windows 10, as I don't have a previous version to test on. Any ideas?

I'm really looking forward to see what I can get out of this new version, particularly the advanced level control.

 

[zeerround]

I tried to use version 1.3 but it keeps on crashing... I started using the default 'play' icon - set values via the web interface and saved an 'ini' file. I made a couple of tweaks to the saved ini file (see attached), but when I then try to run the conversion, it crashes every time... I have tried this on two separate computers with the same result - both are running Windows 10, as I don't have a previous version to test on. Any ideas?

I'm really looking forward to see what I can get out of this new version, particularly the advanced level control.

Yup, congrats, you found the first bug in 1.3. I probably won't have time to squash it today, but you can get around it by editing your ini to change the center width to 0.1, instead of 0.0.

This bug won't affect others that are not setting center width to 0 (for quad) so I would still recommend folks to upgrade now.

 

[Eye of Horus]

Yes, I did !
FWIW I used a FLAC as input.

 

[zeerround]

OK it does look like I uploaded the wrong ini file. Let me find the correct one. My bad.

The one I uploaded has the default widths (with norm only gains and the limiter) so goes to show why the default widths are not appropriate for this song.

 

[zeerround]

Ok I updated the original post with the correct ini file.

 

[Eye of Horus]

Ok I updated the original post with the correct ini file.

I will give it another shot ! Thanks !

 

[Eye of Horus]

The final results with the fixed .ini :



There is quite an improvement on the quality o the sound of the rears, the soundfield (see the values) is still not so good.........
Especially the values of the rears compared to the fronts are odd......
Is it possible to make an .ini where C= +3 compared to L and R, LFE = -3 compared to L and R and the rears are -1 Db compared to R and L .
These are the settings I use on all my conversions. I (tried) adjusted the .ini, but couldn't get these values into the output.
(In 1.2 I simply added -E-3 -R-1, that didn't work either with the .ini file present)

10 minutes after posting this I tried another song from TOAFC and the results were similar..... rears louder than the fronts.
Then I decided to listen to the fronts only and...........the artefacts were hearable in the fronts and the rears were quite clean !?
It looks as if the fronts and rears changed position........


10 minutes later :

I did another test :

Changed usezag from normonly to yes in your .ini
Changed matchlevel to no
Added my wanted levels

And it came out almost as I wanted !

RMS/PEAK
L+R : -22.2/-0.2
C -19.5/-0.2 >>>>>>>> should be -19.2
LFE -23.9/-0.3 >>>>> should be -25.2
Ls+Rs : -23.2/-0.2

The differences in the output versus what it should be are probably caused by the limiter.
But the difference is so small, it's acceptable.
There are still artefacts hearable in the fronts. Some kind of gentle swooshing sound is hearable.

But all in all a major improvement compared to 1.0 to 1.2 versions. EXCELLENT !

 

[zeerround]

Yes, normonly gains ignores the channel to channel output settings in ZAG. Normonly just sets all the gains the same, such that the maximum peak in any channel is equal to your "Max Output" setting (or the original stereo max peak if "matchlevels" is used.

Another way to say that is normonly preserves whatever channel to channel levels come into Zag, as a result of your ArcTan or Slice gains and other settings like widths. So if you want to set channel to channel levels by ear, you could use specweb play and adjust the gain sliders until the channel to channels levels are how you like them, then use normonly gains to preserve them.

If you want the channel to channel volumes to always have a certain ratio (like the old spread sheet), then set ZAG equal to yes and your channel to channel settings will be honored and applied. In this case the levels might be different than what you set by ear with SpecWeb Play, using gain sliders.

If you want to use ZAG set to yes, AND you want to match what you hear in SpecWeb Play, then you should use the meters to calculate the difference between C and Front, Rears and Fronts, etc., and enter those values into ZAG.

Remember that the ZAG overall output level is ignored if using matchlevel.

I hope that's all clear.

 

[zeerround]

Yes, normonly gains ignores the channel to channel output settings in ZAG. Normonly just sets all the gains the same, such that the maximum peak in any channel is equal to your "Max Output" setting (or the original stereo max peak if "matchlevels" is used.

Another way to say that is normonly preserves whatever channel to channel levels come into Zag, as a result of your ArcTan or Slice gains and other settings like widths. So if you want to set channel to channel levels by ear, you could use specweb play and adjust the gain sliders until the channel to channels levels are how you like them, then use normonly gains to preserve them.

If you want the channel to channel volumes to always have a certain ratio (like the old spread sheet), then set ZAG equal to yes and your channel to channel settings will be honored and applied. In this case the levels might be different than what you set by ear with SpecWeb Play, using gain sliders.

If you want to use ZAG set to yes, AND you want to match what you hear in SpecWeb Play, then you should use the meters to calculate the difference between C and Front, Rears and Fronts, etc., and enter those values into ZAG.

Remember that the ZAG overall output level is ignored if using matchlevel.

I hope that's all clear.

 

[Eye of Horus]

Yes, that's all clear ;-)
Still experimenting, this time with more Rock songs....
Same .ini file with my own adjustments. During the years I kind of ignored the different RMS levels between songs on one CD.
That's also why I don't use the spreadsheet anymore ;-)

But despite your explanation : can you tell me why the settings in your ini file delivered such strange values..... (picture of SPECWEB) in my previous post ?
FWIW the song was TOAFC part 1.
The more information, the better my adjustments can be......

 

[zeerround]

Yes, that's all clear ;-)
Still experimenting, this time with more Rock songs....
Same .ini file with my own adjustments. During the years I kind of ignored the different RMS levels between songs on one CD.
That's also why I don't use the spreadsheet anymore ;-)

But despite your explanation : can you tell me why the settings in your ini file delivered such strange values..... (picture of SPECWEB) in my previous post ?
FWIW the song was TOAFC part 1.
The more information, the better my adjustments can be......

OK, let's go back to what started this sub thread, and why I created that ini file. User GOS wondered why SpecWeb results from the "TOAFC" album were strange, with only echos and stuff in the rears. I originally replied (from memory) that he probably needed to adjust the widths to push more of the sound to the rears (original must be mixed with little sound at the left and right edges, more sound toward the center, than your typical stereo track).

Later, you replied about how this was your favorite album, etc., which reminded me again about this corner case (I'm keep track of all "strangeness" mentioned in this thread and using the mentioned songs as tests for improvements to SpecWeb) and made me want to test my theory about changing the widths. I did that, creating an ini file with tighter widths in C and fronts, which meant that more of the sound went to the rears (larger rear width).

Unfortunately, I posted the wrong ini file. The ini file I posted had the DEFAULT width settings, with results that actually proved my theory. This album was mixed with little sound at the extreme left and right, so that with normonly gains (preserving the channel to channel volumes coming out of SpecWeb pre ZAG) the rears were very quiet and only contained few sounds, as the original poster stated.

If ZAG had been set to yes in that ini file, or the default ini settings were used, the rears would be louder (probably too loud for what they contain, in comparison to the original stereo) but would still contain only "echos and stuff".

The correct ini file had C width = 20, frontwidth = 37, vs. the default values of 75 and 90.

By the way, I'm not saying that the settings in the "correct" ini file are the absolute best for TOAFC. As I originally posted I did that quick and dirty (at work, on cheap PC surround speakers) just to demonstrate the concept of pushing more sound to the rears via width settings.

Those of you that put lots of effort and love into each and every stereo to surround conversion are always going to come out with better results than I do, as my personal focus is more on making the tool than using the tool (plus my significant other goes nuts when I play the same song over and over ;0). I will always depend on input from users like you as to optimal default settings and optimal settings for a given song, etc. so keep the constructive feedback on SpecWeb coming.

Thanks,
Z