All times are UTC+02:00




Post new topic  Reply to topic  [ 719 posts ]  Go to page Previous 112 13 14 15 1672 Next
Author Message
 Post subject: Re: Stereo Tool 6.00
PostPosted: Mon Jan 03, 2011 11:38 am 
User avatar

Joined: Tue Mar 17, 2009 2:56 pm
Posts: 4229
Hello!

I checked latest beta and punch is really different. I am still not sure is it better or worse, sometimes i like it and sometimes not. Maybe threshold is little bit higher, still need to test.

I am still don't like MB behavior and 1st three bands joined. Is it possible to join only 1st two (512 mode)? Should be much better!

I hear this artifacts from Phase rotator - very clear when high punch is present, and yes much less when recommended settings are used. I have one question about about Phase rotator from long time ago and maybe this will help for 512 and/or other modes too.
This is 60Hz square tone and StereoTool PhaseRotator ON only:
Image

And This is Same 60Hz square from BA:
Image

As i can see in StereoTool only one "side" is processed by phase rotator. Question is, what will happen if both sides are "done" same as this one.
Maybe will be less needed to use Phase Rotator and then to lead to artifacts free, for 512 mode.


Top
   
 Post subject: Re: Stereo Tool 6.00
PostPosted: Mon Jan 03, 2011 4:15 pm 
Site Admin
User avatar

Joined: Mon Mar 17, 2008 1:40 am
Posts: 11425
Hi Bojcha,

About Punch: Just let me know, I can also create some in-between behavior where I use the old-style Punch type when there's not much loud bass present, sliding to the new one when there is more loud bass (since I had to change it only to get rid of artifacts from deep bass sounds).

About phase rotation: As you've seen with the difficult trumpet sounds in the James Last song, apparently the current phase rotation is very effective (you reported it as being more effective than BA's). The issue is not so much that it causes artifacts at lower latencies (I have gotten it to work nearly artifact-free at latency 512), but that it also affects the 'window' function that I'm using.

Image
Hann window


Image
Hamming window


See http://en.wikipedia.org/wiki/Window_function for more info.


For higher latencies I use a squared Hann window, for lower I use a squared Hamming window - which appears to be needed to reduce certain Loudness artifacts (although it might be useful to check if I can get rid of them another way).


What happens in phase rotation is that certain parts of the audio are moved around in time a bit (say a few samples). This has 2 consequences:
- If I ONLY use Phase Rotation: The 'click' at the start/end of the audio (Hamming window) is moved to different locations for different frequencies! The 2nd filter that cuts off the remaining audio doesn't fully remove it, leaving those clicking sounds.
- Worse: Audio moves to parts of the window where nearly no audio should remain - so for the final limiter and Loudness it appears that the original audio was really loud. Then it tries to cut this, leading to annoying 'bumps' in the waveform, which are especially noticeable for low frequencies (at high frequencies you might hear a very low frequency appear).

I think the only solution for this is to move the phase rotation filter to BEFORE the place where I start working with audio windows. But that requires a very different (0 latency) implementation of the filter - and I'm currently reading into how such filters (in this case I need an 'allpass filter') work.

It seems to be easy enough to get _some_ phase effects, but I want to approach what I have now.


Oh, and one more important thing: I explicitly chose to let the filter move 'forward' only, because of the following:
- Moving FORWARD generally has very little effect on how sounds sound.
- Moving BACKWARD has far more impact: It causes high sounds to be played before lower sounds; clicks hence sound like a 'chok' sound (no idea how to explain this, but you hear the highs before the bass, which can be a nice effect but it's definitely very different from the input signal, and that's not the intention of phase rotation.


Top
   
 Post subject: Re: Stereo Tool 6.00
PostPosted: Mon Jan 03, 2011 8:02 pm 
User avatar

Joined: Tue Mar 17, 2009 2:56 pm
Posts: 4229
I ckecked that Wiki page .. and i just can watch. I must admit that just do not fully understand those things.
But i can hear and measure some basic things, for example songs like this from James Last.
I can adjust Phase Rotator to be ideal for that song, but it's not ideal for other. If i load other track i must re-tweak Phase Rotator a bit to be ideal again.

In other hand, side-effects from Phase Rotator are really minimal. If i compare sound with Orban, where sound is horrible and all bad things are come from Phase Rotator there, ST's Phase Rotator is much much better.
But, now there is need for "Low Latency" i hope you'll find solution for LowLatency modes.

What about multuband? To me that is first thing i don't like in 512 mode.


Top
   
 Post subject: Re: Stereo Tool 6.00
PostPosted: Mon Jan 03, 2011 10:47 pm 
User avatar

Joined: Mon Jun 07, 2010 1:15 am
Posts: 99
You guys are awesome with this, I want to learn more about sound engineering just so I can help out with the dev work. xD

Does the Phase Rotation and Punch problems appear in the 4096 setting? Or is this all just for 512.

_________________
:mrgreen: <-- I like this guy!


Top
   
 Post subject: Re: Stereo Tool 6.00
PostPosted: Mon Jan 03, 2011 11:16 pm 
Site Admin
User avatar

Joined: Mon Mar 17, 2008 1:40 am
Posts: 11425
Bojcha and other who might be interested: This page may explain some things in a simpler way: http://www.katjaas.nl/FFTwindow/FFTwind ... ering.html

Especially check the 2nd and 3rd image (including their explanation), and then the image at the center of the page, which looks a lot like the Stereo Tool artifacts of much older ST versions.

The 1st image is made by multiplying a simple sine wave by the Hann window ( http://www.katjaas.nl/FFTwindow/FFTwindow.html ). The 2nd image is what you get after removing the lows (and highs): The sides are not 0 anymore! And that's the problem.

In Stereo Tool, Final Limiter or Loudness will 'cut' these sides (they appear to be caused by extremely loud sound at that location). And that causes artifacts.


Top
   
 Post subject: Re: Stereo Tool 6.00
PostPosted: Tue Jan 04, 2011 12:34 am 
Site Admin
User avatar

Joined: Mon Mar 17, 2008 1:40 am
Posts: 11425
About Multiband:

I've just done a test where I've listened at what difference level artifacts become noticeable.

Band 1 is not used in 512 sample latency mode -> ignored.
Band 2 vs. 3, 3 vs 4 and 4 vs 5: around 1.4-1.5 times the other value becomes really noticeable. Smaller differences give less artifacts. I don't know yet if there are also artifacts above band 5 (actually I didn't expect them to be so constant).

So it looks like I can use a constant maximum difference value between 2 bands.

So basically what I need to do is make sure that neighboring bands never get more than this apart.

Currently I have encoded 1.125 times for the difference between band 2 and 3 @ latency 512, 1.25 at latency 1024, 1.75 for latency 2048 or 4096 (higher values are used for other bands). I think I should make this configurable - and a lot higher too. Some mild artifacts are probably better than a dysfunctional Multiband filter.


Top
   
 Post subject: Re: Stereo Tool 6.00
PostPosted: Tue Jan 04, 2011 2:06 am 
Site Admin
User avatar

Joined: Mon Mar 17, 2008 1:40 am
Posts: 11425
I've added a slider to Multiband to determine the steepness of the filter - hence also the artifact level.

Some things are not working yet:
- Equalizer. Don't use it! Otherwise you'll still get artifacts. (I think that's also the case in all older versions by the way!)
- In some cases soft sounds may be reduced too much. (This is probably very rare).
- Currently I'm using the same steepness level for all bands. That's not necessary, I think I can determine the artifact level per band and then dynamically adjust the level (higher bands have far less artifacts).

So it's definitely not finished yet, but this should already work a lot better than before.

New slider is set to 1.50 for 512 sample latency; for higher latencies you can (and should!) set it much higher. Again, when I write something to determine the expected latency level, this can be done dynamically automatically.

Building now.


Top
   
 Post subject: Re: Stereo Tool 6.00
PostPosted: Tue Jan 04, 2011 2:58 am 
Site Admin
User avatar

Joined: Mon Mar 17, 2008 1:40 am
Posts: 11425
BETA601-003:
Winamp DSP plugin: http://www.stereotool.com/download/dsp_ ... 01-003.exe
Stand alone version: http://www.stereotool.com/download/ster ... 01-003.exe
VST version: http://www.stereotool.com/download/vst_ ... 01-003.dll
VST version (No SSE2): http://www.stereotool.com/download/vst_ ... 01-003.dll
Command line version: http://www.stereotool.com/download/ster ... 01-003.exe
Linux command line version: http://www.stereotool.com/download/ster ... ETA601-003 [not available]
Linux GUI version: http://www.stereotool.com/download/ster ... ETA601-003 [not available]

IMPORTANT: FOR TESTING PURPOSES ONLY - DO NOT USE FOR 'PRODUCTION'. MULTIBAND IS FLAWED IN THIS VERSION - NEEDS SOME UPDATES (see previous post).

But at latency 512 it should be better, especially at lower bands.


I have by now thought of a way to reliably determine the amount of artifacts per band transition - so that will make it easy to fix the behavior.


Top
   
 Post subject: Re: Stereo Tool 6.00
PostPosted: Tue Jan 04, 2011 5:39 am 

Joined: Fri Mar 26, 2010 2:59 am
Posts: 37
Location: Lake Worth, Florida
I just downloaded/installed and am running version 6.00 with WinAmp. Fantastic !!!!!

Thank you Hans.

Everyone, also have a very Happy Stero Tool New Year.


Top
   
 Post subject: Re: Stereo Tool 6.00
PostPosted: Tue Jan 04, 2011 2:16 pm 
Site Admin
User avatar

Joined: Mon Mar 17, 2008 1:40 am
Posts: 11425
Multiband should now function correctly again, at all latency settings.

'Steepness' is now automatically increased if less artifacts are expected. So for higher latencies - and for higher frequency bands - it has far less effect.

If you set it to '0', Multiband behaves as a very strange singleband compressor (all bands follow the band that's reduced the most). Higher values quickly move towards normal Multiband audio. Default is 0.25 - which means that for the bands with the most artifacts (2 and 3 for latency 512), the output level (compression + equalizer!) may not be more than 25% apart.

BETA601-004:
Winamp DSP plugin: http://www.stereotool.com/download/dsp_ ... 01-004.exe
Stand alone version: http://www.stereotool.com/download/ster ... 01-004.exe
VST version: http://www.stereotool.com/download/vst_ ... 01-004.dll
VST version (No SSE2): http://www.stereotool.com/download/vst_ ... 01-004.dll
Command line version: http://www.stereotool.com/download/ster ... 01-004.exe
Linux command line version: http://www.stereotool.com/download/ster ... ETA601-004 [not available]
Linux GUI version: http://www.stereotool.com/download/ster ... ETA601-004 [not available]


Top
   
Display posts from previous:  Sort by  
Post new topic  Reply to topic  [ 719 posts ]  Go to page Previous 112 13 14 15 1672 Next

All times are UTC+02:00


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
Powered by phpBB® Forum Software © phpBB Limited