SM2 vs SPECWeb vs PENTEO 16 Pro

QuadraphonicQuad

Help Support QuadraphonicQuad:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
I think I need some help with SW2. I tried playing using the -m4 option but I'm still hearing artefacts, swishing from about the 22 sec mark in the test file

snip...snip

Glenn, can you suggest settings to remove the swishes, drums and guitars from 22 sec mark?

Both upmixes sound great until 22 sec mark, then SpecWeb starts swishing like crazy.

THX

Man that song is the poster child for "swish", a good find. I will add to my torture tracks for trying to eliminate the swish.

I think they used some crazy stereo widener on that track, because in stereo stuff is clearly coming from left of left and right of right, although it doesn't show up as funny on a phase meter. I might try "narrowing" the mix, before spreading it back out with SpecWeb, just to see what happens.

Anyway to answer your question as to settings, I came up with the attached (rename to remove .txt extension), and then you can add in -m4 on the command line if you like.

I'm not saying it's a complete solution, but then I don't think I found the same source as yours (mine is 44.1K 16bits).

Longer version is:

"Swish" is a known artifact of Phase Vocoders, the way Spec and SpecWeb move sound from the time domain to the frequency domain and back. When/If I find a better way (which may just be DSP techniques I don't grasp yet) I will change it and not look back. At one point I completely ported a different method, fast hartley transform vs. fast fourier transform, from "CenterCut" from C# to C++ but it ended sounding exactly the same (the time to frequency and back part, not the "CenterCut" method itself).

"Swish" with phase vocoders happens with "Noisy Transients" like certain cymbal sounds. Normally these are tamed with the upsample --> upmix --> downsample technique but for some songs, like this one, it's not a complete solution.

So, in this case the next thing you can do (if 2x and 4x upsample don't help) is trade off separation for a reduction in artifacts. There are a bunch of controls in Spec/SpecWeb for this. Mainly Adjacent Speaker (mixes in sounds from the speaker on either side, e.g. mixing some RF and LS into RS) but probably more applicable to this case mixing some of the original stereo back into the rear channels, with the LS and RS Blend controls.

As I said, this is trading off artifact reduction for less separation so at the end of the day you might have waveforms that look/sound similar to Penteo, I guess.
 

Attachments

  • SpecWeb.ini.txt
    11 KB · Views: 262
Hey Glenn, First of all thanks again for all you do. I exclusively like to use the slice method in regular spec, mostly because it is very, very clean and it lends itself to nice seperation, As long as I do not try funky processing like expanders and such there are Never any artifacts. Maybe you can somehow incorporate the slice method as an option into specweb. I personally never liked arctan simply due to its high possibility of creating odd artifacts such as that swishing sound, faint echos, and what I call the blanket, which is the sound of playback from under a blanket. Kevin
 
Slice is there in SpecWeb. I just noticed tonight that (at least in "Play") the level is low. I will look into that. In the meantime it can probably be fixed with Pregain:

-Gx = Set pregain to x dB (default 0.0).



Anyway, Slice settings In the ini file:

[method]​
;slice, arctan, arctan+slice​
;​
mode=arctan​
[slice]​
;A Humidity of 1 = 100% wet, or SLICE processed signal​
;A Humidity of 0 = 0% wet or 100% dry signal from the previous stage​
; (or original left and right in the case of stage one).​
;Stage 1 Humidity default 0.9​
stage1humidity=0.9​
;Stage 2 Humidity default 0.95​
stage2humidity=0.95​
;​
;Wrap Rears adds a third stage of separation​
;This has the effect of taking what was the extreme​
;outside of the original stereo field and placing it​
;in both rear speakers, creating a virtual center rear​
;on, off default: off​
wrap=off​
;Stage 3 Humidity default 1.0​
stage3humidity=1.00​
;​

On the command line:

-Mx = SpecWeb Method (0=Slice, 1=ArcTan, 2=ArcTan with Slice Blends).​
-ox = Set Slice Stage One Humidity to x (default 0.9).​
-tx = Set Slice Stage two Humidity to x (default 0.95).​
-wx = Set Slice Wrap Rears to x, 0 = off, 1 = on (default is off).​
-Tx = Set Slice Stage three Humidity to x (default 1.0).​

And there are controls in the Web Interface (and console) during Play.
 
Man that song is the poster child for "swish", a good find. I will add to my torture tracks for trying to eliminate the swish.

Thanks for you detailed response Glenn. I’ll have a play around as you’ve suggested.

I‘ll also checkout the Slice method suggested by Kevin. (Thanks Kev)
 
Upmixing is a chore. How do you decide?

Here is a 30 second section of the stereo tune "Crazy Love" from Poco's LEGEND album. It has very good vocal separation among the front and rear channels. Here are 3 versions. SPECWeb 1.5 (I did this a year ago), SPECWeb 2.0x with -m4, and Penteo 16 Pro using DEFAULT.

After I hear each one, I think "That's it!", then I listen to the next and do the same thing. These are 16/44 so you can play them right here if you have the proper playback on your PC - or just download and then you can look at the files themselves.

https://www.quadraphonicquad.com/Samples/TEST-SW15.wav
https://www.quadraphonicquad.com/Samples/TEST-SWM4.wav
https://www.quadraphonicquad.com/Samples/TEST-PENT.wav
 
Upmixing is a chore. How do you decide?

Here is a 30 second section of the stereo tune "Crazy Love" from Poco's LEGEND album. It has very good vocal separation among the front and rear channels. Here are 3 versions. SPECWeb 1.5 (I did this a year ago), SPECWeb 2.0x with -m4, and Penteo 16 Pro using DEFAULT.

After I hear each one, I think "That's it!", then I listen to the next and do the same thing. These are 16/44 so you can play them right here if you have the proper playback on your PC - or just download and then you can look at the files themselves.

https://www.quadraphonicquad.com/Samples/TEST-SW15.wav
https://www.quadraphonicquad.com/Samples/TEST-SWM4.wav
https://www.quadraphonicquad.com/Samples/TEST-PENT.wav

Yeah, they all sound good to me, just real subtle stuff to consider, so hard to pick a favorite.
One thing I do notice on up-mixes like these, that I'd like confirmation and feedback from others on; is on certain songs during artists singing the letters S or Z, I'm hearing a kind of an "artifact" like crispiness to them. Maybe Glenn could comment on that and recommend something, if others also hear that.

Also have to say I'm absolutely thrilled as a young buck on his first date, to have these tools to convert songs from old quasi two dimensional stereo into a multi-channel dreamland. :smokin
 
Last edited:
Upmixing is a chore. How do you decide?
Yep that's $64k question that evidently doesn't have a real clear cut answer. Part of the answer may not even have to do with decoding quality.

I was rather disapointed when DPL II came out as it really didn't seem to advance in soundfield quality over older methods. So I spent quite some time dveloping my own S2S method & posted it here several years ago on the forum. It involves pre-synthesis & other wavform editing prior to decoding, then followed an RM type decoding with high seperation enhancement & finish off after decoding with level, EQ, panning, etc. Of all PC methods I've used it performs the best but is very labor intensive. So when SpecWeb arrived ( I never could wrap my head around all that Plouge Bidule stuff) it took me 90% there compared to my previous method. And you can still do pre-synth & post decode editing with SpecWeb if you want to go to the trouble. If I wanted to bring a surround sound chunk of music over to someones house that only had DPL II or Atmos upmix or whatever I would use SpecWeb. For something really important to me I would use my old way, for something I would want to listen to many times in the future.

And as for stuff like Penteo I've had my disappointments. One of course, as I mentioned before, was not getting the trial Penteo to work at all a few years back. I also purchased dts Neural Upmix & found out that AA3 it couldn't handle the VST 3. Besides the frustration this stuff is too expensive to impulse buy. So, then another good reason for this note comparing it all.

And when I think I've ran across most all of the plug in FX upmixers & stuff I find this just today. Regular price $149 on sale for $30. Who can't afford that? But I can't imagine it providing any superior benefit to SpecWeb.

My own answer to which is best would be hardware decoding, that is, the SM v2. If you balance the factors of ease of use, quality of soundfield, artifact free & cost I think the SM is a fair dinkum value. I have a benefit that I don't think anyone else has here with the SM. In another note I posted about modifying a Chase RLC-1 to use as a front end for the SM. It allows for 4 inputs, bass/treble, L/R level balance & phase balance. I can not over state what a benifit it adds to be able fine tune those last 2 items, by remote control from the sweet spot, & then just let the SM v2 work it's magic. My need to do S2S on the PC has dropped almost to zero. Now on the other hand you must have analog in/out capability on your hardware to use the SM. If someone has only HDMI that's a deal breaker. So then, back to the PC for surround upmixing.

Is anyone going to try the Waves surround upmix & compare?
 
FYI I already own the Waves upmix. It came as part of a bundle (of now outdated in that it only goes to 5.1 and doesn't support Pro tools modern plugin format) surround VSTs. Honestly don't recall what the upmix sounds like. Apparently it didn't amaze me at the time.

...

A lot of these older things are starting to die off. People that made good VSTs moving on to other stuff, and plugin & host formats changing to 64bits only (Like Plogue Bidule).

...

I spent some time looking at hardware platforms last week, and short of Raspberry Pi, I can see why any hardware based solution is going to have to sell for over $300. Unless you are designing your own circuit board, that's the minimum that the DSP and A/D D/A hardware is going to cost.

You could potentially save some money with HDMI, but while the audio breakout stuff is out there, re-embedding audio means a HDMI license and hardware I haven't seen (There is high end audio hardware using HDMI cables for digital audio but it is NOT HDMI audio format).

Raspberry Pi is not really DSP, but probably the least expensive way to go and maybe the least amount of work to port something like Spec/SpecWeb to. Quality would probably be the same as SpecWeb and since you CAN do SpecWeb "live on the fly", I'm not sure it makes any sense to do a standalone hardware version anyway (assuming you have a windows computer).
 
Is anyone going to try the Waves surround upmix & compare?

yea Sonik that was one that the older upmixers used, like area51 and toup, we used to kick that process around on the extinct Doom9 site, you can get pretty good results if you work your ass off at it. However, Spec outperforms it in every way.

I didn't know that slice was available on the web version, you guys should try that one, at least with the full plogue version I find it to be much cleaner. However, if you are one of those that feel vocals Must be in the center channel then Arctan is for you. in which case I highly recommend you read the guide that Jorge wrote entitled "how to win with arctan" because it can produce a lot of funkiness.
 
I always like to mention that a little out of phase pre-synthesis really expands the sound stage in QS. As I'm sure your familiar with the difference between the basic QS vs Synthesis mode on a Sansui, or the stereo surround mode on a Tate, you know what I'm talking about. If you get the chance open the original file in Sound Forge & do -10db or -7dB opposite phase blending & try this test again. I'm sure it would be very illuminating

I got a free offer to get iZotope Imager plug-in. Its widening/narrowing a stereo image. Does this look like its pre-synthesis Sonik Wiz? I'm going to try it over the week end.

https://www.izotope.com/en/products/ozone-imager.html
There's a short video on its use on the link above. Looks interesting.
 
I got a free offer to get iZotope Imager plug-in. Its widening/narrowing a stereo image. Does this look like its pre-synthesis Sonik Wiz? I'm going to try it over the week end.

https://www.izotope.com/en/products/ozone-imager.html
There's a short video on its use on the link above. Looks interesting.
So, looks like it can go through Reaper, and is "not" just a trial time limit... is that right Garry?

Edit* - I'm not seeing a "sign up" page, just more sales info...???
 
I got a free offer to get iZotope Imager plug-in. Its widening/narrowing a stereo image. Does this look like its pre-synthesis Sonik Wiz? I'm going to try it over the week end.

https://www.izotope.com/en/products/ozone-imager.html
There's a short video on its use on the link above. Looks interesting.

Yes it looks like it's applicable for stereo pre-synth. Probably a bit of over kill with graphics & other functions but I'm sure it would be fun to play with!

Internally it's probably performing the function of summing both chs to create a Mid signal & differenceing the 2 chs to create a Side signal. The slider adjusts the balance between the 2 & then they are matrixed again to get your 2 stereo chs back. Like the AA Expand function the number scale seems arbitrary & you would have to experiment a bit to find a basic starting point. With QS/Surround Master decoder you could input only the left ch & adjust the slider for strongest output left back. For your Denon & Arcam I am not familiar with their functions so I can't advise specifically. But if you can play with this real time I bet you'll find the sweet spot fast!

Keep us informed!
 
So, looks like it can go through Reaper, and is "not" just a trial time limit... is that right Garry?

Edit* - I'm not seeing a "sign up" page, just more sales info...???

I bought a Black Friday special to get the izotope declicker so I’m a registered user. I got an email this morning with the free offer to get the Ozone Imager. Not sure if it’s free for non-users.

There’s still specials and you can get the Declicker for $49. I’ll post a link.

Yes, these are all VST3 plugins that work in a Reaper
 
I bought a Black Friday special to get the izotope declicker so I’m a registered user. I got an email this morning with the free offer to get the Ozone Imager. Not sure if it’s free for non-users.

There’s still specials and you can get the Declicker for $49. I’ll post a link.

Yes, these are all VST3 plugins that work in a Reaper
Thanks Garry, also do you know if that Declicker includes "DeCrackle"?
 
Yes it does

Are you sure about that? According to the website, "DeCrackle" is only available in the RX "Standard" bundle and above. I went for the Black Friday deal (which includes the base-level RX "Elements" bundle) and was bummed to find you only get "DeClick", "Voice DeNoise", and "DeClip".

decrackle.jpg


For those who do conversions from vinyl sources - the "DeCrackle" is definitely worth checking out. I used it on my last couple SQ LP needle-drops and the results are stunning.
 
Back
Top