Stereo Tool
https://forums.stereotool.com/

Stereo Tool 6.10
https://forums.stereotool.com/viewtopic.php?t=3065		 Page 89 of 213
Author:  hvz [ Thu Jul 07, 2011 2:04 am ]
Post subject:  Re: Stereo Tool 6.10
Quote:
Quote:
The current clipper is doing some things that it shouldn't. Most importantly I think I have to completely get rid of the Clean/Punch filter, and make sure that the punch is maintained using other methods. The Clean/Punch filter - when not set to Clean - causes the Loudness filter to behave partly as a compressor, which is bad, and it makes artifacts such as voice vibrations, damaged 'S' sounds and volume drops caused by loud highs much worse. I have some new ideas which I'll test soon.
Exactly .. every word!
Yeah :-) And also important: The simpler I make my clipper, the easier it will be to convert it to a composite clipper (actually that should take about a day once I have my L/R clipper ready using my new ideas).

Currently I'm doing a lot of things simultaneously:
- Lowering frequencies that cause issues (for example, loud bass if 'S' sounds are present, or loud 'S' sounds if they push other frequencies down too much, ...)
- Clipping
- Filtering
- Increasing punch

Well this is just getting far too complex. So I want to split it into simple steps. When that's done, composite clipping should be only slightly more difficult than normal clipping, and it won't cause too much CPU load (it would if I would do it in my current code).
Author:  Bojcha [ Thu Jul 07, 2011 2:12 am ]
Post subject:  Re: Stereo Tool 6.10
I can't talk how you will do it .. because i don't know programming...

But there is another potentional problem.
After research with Romke.. you can't use L/R clipper in same time with Composite CLipper. So question is how would be using in same time FM and Normal Output.

In Omnia.9 there is whole separate processing for FM and Normal output/s... Meaning separate pre-emphasis and separate clippers (?) So that is only soulution for that configuration - using both FM and Normal Out) (?)
Author:  hvz [ Thu Jul 07, 2011 2:47 am ]
Post subject:  Re: Stereo Tool 6.10
Quote:
I can't talk how you will do it .. because i don't know programming...

But there is another potentional problem.
After research with Romke.. you can't use L/R clipper in same time with Composite CLipper. So question is how would be using in same time FM and Normal Output.

In Omnia.9 there is whole separate processing for FM and Normal output/s... Meaning separate pre-emphasis and separate clippers (?) So that is only soulution for that configuration - using both FM and Normal Out) (?)
Hm, I guess you're right. On the other hand, if I can split things the way I want, only the last stage (the clipping itself, which is relatively simple) would have to be done twice.
Author:  Brian [ Thu Jul 07, 2011 5:37 am ]
Post subject:  Re: Stereo Tool 6.10
All of this composite clipping is FM-only, right?

Also, is there any feeling how a redesigned non-FM loudness section would be on CPU load? As I've previously mentioned, the CPU load of loudness is currently the same load as the Declipper as viewed from my system. I can use Natural Dynamics (just barely) if I disable loudness, but I lose all the benefits of the current loudness section...
Author:  hvz [ Thu Jul 07, 2011 5:39 am ]
Post subject:  Re: Stereo Tool 6.10
Quote:
All of this composite clipping is FM-only, right?

Also, is there any feeling how a redesigned non-FM loudness section would be on CPU load? As I've previously mentioned, the CPU load of loudness is currently the same load as the Declipper as viewed from my system. I can use Natural Dynamics (just barely) if I disable loudness, but I lose all the benefits of the current loudness section...
I *hope* it will be lower. The Punch thing causes lots of issues that need to be fixed, which takes (a lot of) extra effort. So if I can get rid of that, it should solve a lot of issues.
Author:  Brian [ Thu Jul 07, 2011 5:42 am ]
Post subject:  Re: Stereo Tool 6.10
Quote:
Quote:
All of this composite clipping is FM-only, right?

Also, is there any feeling how a redesigned non-FM loudness section would be on CPU load? As I've previously mentioned, the CPU load of loudness is currently the same load as the Declipper as viewed from my system. I can use Natural Dynamics (just barely) if I disable loudness, but I lose all the benefits of the current loudness section...
I *hope* it will be lower. The Punch thing causes lots of issues that need to be fixed, which takes (a lot of) extra effort. So if I can get rid of that, it should solve a lot of issues.
OK. I see that your to-do list still has optimizations for Natural Dynamics as well, so I'm hoping that when all is finished, I can squeeze in both loudness and natural dynamics at about 75-80% total load... It'll be a bit sluggish, but tolerable.

Hmmm.... One other thing I just noticed. Noise Gate has a very high CPU load... 26-28%... Hmmm...very strange. If I enable just Noise Gate, the CPU usage goes up to 30%. If I enable Loudness, I also get to around 30%, but if I then enable Noise Gate along with loudness, usage only goes up to 40-45% :?:
Author:  hvz [ Thu Jul 07, 2011 7:20 am ]
Post subject:  Re: Stereo Tool 6.10
That's because the CPU usage values from Task Manager are extremely unreliable - 20% jumps in the reported values with nearly no change in actual CPU usage are not only possible but happen very frequently.
Author:  Brian [ Thu Jul 07, 2011 8:56 am ]
Post subject:  Re: Stereo Tool 6.10
Quote:
That's because the CPU usage values from Task Manager are extremely unreliable - 20% jumps in the reported values with nearly no change in actual CPU usage are not only possible but happen very frequently.
We've been down this path before. At that low of a utilization, no, I will not see system sluggishness. The cummulative effect of everything though, once all the filters interact with each other, then yes, I do see sluggishness.

Do you have any supporting links to discussions from reliable sources in regards to Task Manager and this "unreliable" metric, or is this potentially a perpetuated myth? I've done a quick search on Google, and don't see anything obvious saying "Task Manager's CPU Usage metric is flawed" or anything like that, although I will admit, I have just recently taken 2 pain pills and an anti-anxiety med to help me sleep, and it's kicking in... :|

Have you gone over the code in the Noise Gate to make sure it is optimized? At one time in the past, you had yourself said it was an area that you rarely used. Also, per another post on Jan 13, 2011:
Quote:
The noise gate code is a bit strange and it's easy to make bugs in it. So I guess that's what I did... I'll check it later.

Edit: 9.99 -> 10.00 jump should not make any difference. But if the muddy sound disappears when you touch the slider, it must be the preset loading that does something strange.
Author:  hvz [ Thu Jul 07, 2011 3:08 pm ]
Post subject:  Re: Stereo Tool 6.10
@Brian: I did measurements with Task Manager and VTune (a program that checks which code is executed in the system ~ 10,000 times per second). And with those two I could change a single line of code in Stereo Tool (with no clear relation to performance), which caused 20% jumps in Task Manager but showed no difference at all in VTune.

This was measured with the Winamp plugin. Since I've seen this I measure the performance by letting Winamp write the output to a file, so Stereo Tool processes the data as soon as possible. Again, the two versions with 20% difference in Task Manager if I just let it play the output show NO difference at all using this measurement.

The CPU usage of Noise Gate should be about 70% of that of Multiband, clipping enabled. The strange part is the parameter handling, not the behavior or performance.
Author:  Luke [ Thu Jul 07, 2011 9:07 pm ]
Post subject:  Re: Stereo Tool 6.10
Quote:
@Brian: I did measurements with Task Manager and VTune (a program that checks which code is executed in the system ~ 10,000 times per second). And with those two I could change a single line of code in Stereo Tool (with no clear relation to performance), which caused 20% jumps in Task Manager but showed no difference at all in VTune.

This was measured with the Winamp plugin. Since I've seen this I measure the performance by letting Winamp write the output to a file, so Stereo Tool processes the data as soon as possible. Again, the two versions with 20% difference in Task Manager if I just let it play the output show NO difference at all using this measurement.

The CPU usage of Noise Gate should be about 70% of that of Multiband, clipping enabled. The strange part is the parameter handling, not the behavior or performance.

Hans I must say that the noise gate is a very useful feature in stereotool. This thing rocks when playing 70's music, especially intro's which seem to come directly from tape you can mask most of the noise away.
Don't know if it is an easy request but would it be possible to have an indicator in that window so it makes it easier to monitor if the levels have been set correctly?
Page 89 of 213 All times are UTC+02:00
Powered by phpBB® Forum Software © phpBB Limited
https://www.phpbb.com/