Thanks for the confirmation, that will certainly do OK, a bit of control from time never hurts
Of course.
BTW, looked the sources and (if I got it right) looks like you use Butterworth type LP/HP filters for IIR type filtering ... which does maybe not produce the best possible response as shown in attached comparison plot.
Here's Octave/Matlab code for the LR4 vs 4th order Butterworth response comparison I made (these are 4th order filters but one can transfer those to 2nd order sets):
Code:
pkg load signal
pkg load control
clf;
fs = 44100;
fc = 2500;
% Butterworth filters
[Lb, La] = butter (4, fc/(fs/2));
ButterLP4 = tf(Lb, La, 1/fs);
[Hb, Ha] = butter (4, fc/(fs/2), "high");
ButterHP4 = tf(Hb, Ha, 1/fs);
% Linkwitz-Riley filters
[b, a] = LR4_HP4(fs, fc);
HP4 = tf(b, a, 1/fs);
[b, a] = LR4_LP4(fs, fc);
LP4 = tf(b, a, 1/fs);
% sum response
LR4 = HP4 + LP4;
ButterLR4 = ButterLP4 + ButterHP4;
% coefficients
[b, a, T] = tfdata(LR4, 'v')
[b, a, T] = tfdata(ButterLR4, 'v')
nf = logspace(0, 5, fs/2);
figure(1);
[mag, pha] = bode(LP4,2*pi*nf);
semilogx(nf, 20*log10(abs(mag)), 'color', 'r', 'linewidth', 2, 'linestyle', '-');
hold on;
%semilogx(nf, pha, 'color', 'r','linewidth', 2 , 'linestyle', '--');
hold on;
[mag, pha] = bode(HP4,2*pi*nf);
semilogx(nf, 20*log10(abs(mag)), 'color', 'r', 'linewidth', 2, 'linestyle', '-');
%semilogx(nf, pha, 'color', 'b','linewidth', 2 , 'linestyle', '--');
hold on;
[mag, pha] = bode(LR4,2*pi*nf);
semilogx(nf, 20*log10(abs(mag)), 'color', 'g', 'linewidth', 2, 'linestyle', '-');
%semilogx(nf, pha, 'color', 'g','linewidth', 2 , 'linestyle', '--');
hold on;
[mag, pha] = bode(ButterLP4,2*pi*nf);
semilogx(nf, 20*log10(abs(mag)), 'color', 'k', 'linewidth', 2, 'linestyle', '--');
%semilogx(nf, pha, 'color', 'g','linewidth', k , 'linestyle', '--');
hold on;
[mag, pha] = bode(ButterHP4,2*pi*nf);
semilogx(nf, 20*log10(abs(mag)), 'color', 'k', 'linewidth', 2, 'linestyle', '--');
%semilogx(nf, pha, 'color', 'g','linewidth', k , 'linestyle', '--');
hold on;
[mag, pha] = bode(ButterLR4,2*pi*nf);
semilogx(nf, 20*log10(abs(mag)), 'color', 'b', 'linewidth', 2, 'linestyle', '--');
%semilogx(nf, pha, 'color', 'g','linewidth', k , 'linestyle', '--');
legend('LR4 LP', 'LR4 HP', 'LR4 SUM', 'Butterworth4 LP','Butterworth4 HP','Butterworth4 SUM', 'location', 'southwest');
axis([10 fs/2 -30 1]);
grid on;% Linkwitz-Riley
Code:
function [b, a] = LR4_LP4(fs, fc)
wc=2*pi*fc;
wc2=wc*wc;
wc3=wc2*wc;
wc4=wc2*wc2;
k=wc/tan(pi*fc/fs);
k2=k*k;
k3=k2*k;
k4=k2*k2;
sqrt2=sqrt(2);
sq_tmp1=sqrt2*wc3*k;
sq_tmp2=sqrt2*wc*k3;
a_tmp=4*wc2*k2+2*sq_tmp1+k4+2*sq_tmp2+wc4;
b0 = 1;
b1=(4*(wc4+sq_tmp1-k4-sq_tmp2))/a_tmp;
b2=(6*wc4-8*wc2*k2+6*k4)/a_tmp;
b3=(4*(wc4-sq_tmp1+sq_tmp2-k4))/a_tmp;
b4=(k4-2*sq_tmp1+wc4-2*sq_tmp2+4*wc2*k2)/a_tmp;
a0=k4/a_tmp;
a1=-4*k4/a_tmp;
a2=6*k4/a_tmp;
a3=a1;
a4=a0;
a = [b0 b1 b2 b3 b4];
b = [a0 a1 a2 a3 a4];
endfunction
Code:
function [b, a] = LR4_HP4(fs, fc)
wc=2*pi*fc;
wc2=wc*wc;
wc3=wc2*wc;
wc4=wc2*wc2;
k=wc/tan(pi*fc/fs);
k2=k*k;
k3=k2*k;
k4=k2*k2;
sqrt2=sqrt(2);
sq_tmp1=sqrt2*wc3*k;
sq_tmp2=sqrt2*wc*k3;
a_tmp=4*wc2*k2+2*sq_tmp1+k4+2*sq_tmp2+wc4;
b0 = 1;
b1=(4*(wc4+sq_tmp1-k4-sq_tmp2))/a_tmp;
b2=(6*wc4-8*wc2*k2+6*k4)/a_tmp;
b3=(4*(wc4-sq_tmp1+sq_tmp2-k4))/a_tmp;
b4=(k4-2*sq_tmp1+wc4-2*sq_tmp2+4*wc2*k2)/a_tmp;
a0=wc4/a_tmp;
a1=4*wc4/a_tmp;
a2=6*wc4/a_tmp;
a3=a1;
a4=a0;
a = [b0 b1 b2 b3 b4];
b = [a0 a1 a2 a3 a4];
endfunctionAttachments
Running new version from the RAW branch files.
What I am observing today and yesterday is fairly high output when mixed to mono.
This is what I used to mix to mono with Brutefir:
pcm.monocard{
slave.pcm "hw:Loopback,DEV=0"
slave.channels 2
type route
ttable {
# Copy both input channels to output channel 0 (Left).
0.0 0.5
1.0 0.5
# Copy both input to output channel 1 (Right).
0.1 0.5
1.1 0.5
}
}
pcm.!default monocard
It has to be set to 0.5 output, otherwise way to high output.
Does the Camilla config when mixing to mono also do this?
Sry... i didn't make it yesterday
I've got this crappy new cheap dac working, but only with S16_LE, but nevermind that now, this is just for testing.
So all my "Alsa" configs are working, also when showing the .py script.
The playback are full of glitches, and perhaps i have to adjust the dac/squeezelite settings, BUT i would also like to try the output rate control, what do i have to "git" for trying that ?
Jesper.
Attachments
Sry... i didn't make it yesterday
I've got this crappy new cheap dac working, but only with S16_LE, but nevermind that now, this is just for testing.
So all my "Alsa" configs are working, also when showing the .py script.
The playback are full of glitches, and perhaps i have to adjust the dac/squeezelite settings, BUT i would also like to try the output rate control, what do i have to "git" for trying that ?
Jesper.
Regarding CamillaDSP with Rate control feature, it's located under the the branch ''develop''.
Thanks...
I installed this version allready, how do you make it write out the buffer, like this you posted a while ago ?
I am not sure it's working with my .yml file?
Your settings under devices, target_level and adjust_period, are no different than mine, so I think it should..
Okay... The file plays, but do still glitch. but in your'e earlier post the output was like this
I cannot see this on output?
Jesper.
Okay... let's see if Henrik know's about this
I'am just curious to see and adjust the buffers so i can have glitchfree playback perhaps.
Jesper.
Forgot to note that LR4_HP4 filter implementation presented in quoted post has bad LF issue when value of fc is set to less than 1/63 * fs/2 (i.e. 350Hz @ 44.1kHz) ....
Now I think of it, I run squeezelite with default settings. No glitches.
Jesper posted these settings:
squeezelite -n CamillaDAC -o squeeze -a 160 4 1 -b 10000 20000 -C 15 -z
The High- and Lowpass filters have adjustable Q so they can be anything you like. For example to make a 4:th order Butterworth you chain one with Q=0.54 and one with Q=1.31. I'm making a table of the most common types that will be included in the readme for the next version.
The config for the mono-mixer I posted attenuates both channels by 6dB, so should give the same as your BruteFIR config. Do you get clipping warnings?
