Stereo Tool

Yes. Of course the CPU load must be a lot lower then. On the other hand, the filter is so much better that you can use FAR less bands. Which should help a lot.

Fix for the bug in version 019:
Stand alone: http://www.stereotool.com/download/ster ... 04-020.exe
Winamp DSP: http://www.stereotool.com/download/dsp_ ... 04-020.exe
VST: http://www.stereotool.com/download/vst_ ... 04-020.dll

I need to leave now, so I haven't had time yet to check the audio. But this thing behaves exactly according to the descriptions of other compressors - it calculates an envelope and performs attack and release on that envelope - and the 018 version does not. Have been doing the financial adminstration this afternoon, I *really* hate that...

Still release changes with different threshold.

There are also some effects that do not seem right.

- If you set the ratio to 1:1 value, should not apply any gain reduction regardless of any adjustments, and if you are doing.

- If you set the ratio to 1000:1, the VU accuses gain reduction to lower the threshold to less than -12 dB, even if the reduction is audible.

- If you keep the Threshold to -12 dB and the ratio is changed to a very low 1.50:1 (for example), the VU accuses greater gain reduction with a ratio of 1000:1, although perceived audibly less reduction.

- It seems that there is an abrupt change of behavior around the -10 to -12 dB Threshold, where higher values ​​there is no activity in the compressor.

Always keeping the same attack = 10 ms, release = 500 ms and a knee = 10 dB

_________________
by GAP
"Less is More" (Bob Katz)

I want to write something off topic, I try do download from lastest link above and I downloaded something strange just setup.exe.

Ok, it is my problem, some plugin in firefox

_________________
Urban Junglist

Tested with ratio 4:1 and 1000:1, feedback and feedforward, in each case compared threshold -12 and -50 dB. In all cases the two shapes were completely identical. Are you maybe using Knee? That changes the shape - AND IT SHOULD!

Am I missing something?


Will now check GAP's feedback...

Tested, output is identical to input, tried both feedback and feed forward.
Level ends up at the set threshold, constantly. Looks ok... Of course only when using feedforward. What do you mean by VU, the display?
You mean it shows more difference? Cannot confirm that here so far... Tried both feedforward and feedback.
If you don't use Knee the release doesn't slow down near the threshold so it will very quickly get back to a gain of 1.



I'm really confused now...

@GAP: Will your the doc in your pm now. But maybe a short question. As you seem to know a lot about this stuff.

What's the correct behavior?

Assuming *no* knee:
1. If level drops below threshold, the release happens at normal speed for the amount of volume drop until a gain of 1 is reached, then it stops abruptly.
2. If level drops below threshold, release is done as if the level is equal to threshold - hence it will be slower.

In case '1', Knee would be used to make the abrupt change smooth. Based on what you wrote earlier this makes sense, but Bojcha says that none of the compressors that he ever tested (and I know he tested a whole lot of them) does this except the one from Reaper.

This is correct, when the gain equals 1 Release stops, because it reaches its target.
Confuses me that the Release should be made ​​as if the Threshold level is equal to, at first glance this does not seem right. It must remain as had been calculated.

We can see in the simple example send before:

/ / Dynamic selection: attack or release?
double theta = rms> env? att: rel;

/ / With smoothing capacitor, envelope extraction ...
/ / Here be aware of PIV denormal numbers glitch
env = (1.0 - theta) + theta * rms * env;

/ / The hard knee very easy 1: N compressor
double gain = 1.0;
if (env> threshold)
gain = gain - (env - threshold) * slope;
La rodilla ayuda que el cambio de Ratio de 1:1 a 1:n sea mas suave, principalmente en el Attack por los tiempos mas cortos que se manejan. Pero en el Release no debería ser muy notoria la rodilla, debido a que los tiempos son mayores.

Just going to check this, at night, in the papers. But these are the most inaccurate in all do something different, according to the theory of each.

_________________
by GAP
"Less is More" (Bob Katz)

In this regard, I believe that everything that happens is an error in the indicator display gain reduction (which I called VU = Vumeter).

Since keeping all parameters constant and only varying the ratio control is less gain reduction for high Raios that lower ratios, even with 1:1 ratio gain reduction is, when for that case there should be no knee the gain must equal 1.

But it may be a display bug, not levels.

_________________
by GAP
"Less is More" (Bob Katz)