IvnOs

At this moment i'm trying to make a Q/gain compensation for white noise which is going thru [biquad~] which is switched to bandpass. In other words, i want to have a nearly constant gain of my white noise, while i'm changing Q.

Please, any links on connection of gain and Q, will be very helpful.

Hi.
At first sorry if my Eng is not so good.

Please, any links on connection of gain and Q, will be very helpful.


bp-filter-q-and-gain-compensation

Found some exponential dependence between Q and processed noise gain.

Also i found that gain is also dropping down with lowering the filter frequency.

Here some numbers for white noise thru [biquad~] set to

For now my brain is becoming to collapse... Any suggestions about how to make an expression for gain compensation will be very helpful.

Also here is patch which i used to calculate values.

Found some exponential dependence between Q and processed noise gain.
Also i found that gain is also dropping down with lowering the filter frequency.

Here some numbers for white noise thru [biquad~] set to
13kHz
Q __ Lvl(average peak for 3000 ms)
0.1 __ 1.5
0.5 __ 1.17
1 __ 1.05
10 __ 0.5
100 __ 0.15
1000 __ 0.05

3kHz
Q __ Lvl
0.1 __ 1.4
0.5 __ 1.1
1 __ 0.9
10 __ 0.33
100 __ 0.1
1000 __ 0.025

500Hz
Q __ Lvl
0.1 __ 0.93
0.5 __ 0.6
1 __ 0.4
10 __ 0.13
100 __ 0.04
1000 __ 0.01

Also here is patch which i used to calculate values.


I found solution. Becoming so stupid sometimes =)

[peakamp~] catches peak amplitude after [biquad~], after that 1. is divided on recived value.

And the finishing object [*~] just multiplying dry signal with processed peak value.

For now the problem is that high time values of [peakamp~] taking a long waiting time before correct amplitude after shifting Q value, which means that signal becoming to low or to loud for some time. Also lower time values making some kind of jitting amplitude modulation.

Also, i realy can't understand how amplitude of noise can be more then input amplitude if we are cutting frequencies?

I found solution. Becoming so stupid sometimes =)
But it is not yet so good as it is.
For now it is looking like that

[peakamp~] catches peak amplitude after [biquad~], after that 1. is divided on recived value.
And the finishing object [*~] just multiplying dry signal with processed peak value.

BP filter Q and Gain compensation