How do I plot a spectrogram of a real time plugin?

8 views (last 30 days)
I managed to program a plugin that does a bit of reverb and a low pass filter. The plugin itself just shows me the spectrum of a signal during the time it is played but I need the entire spectrogram of that signal. Is there an option to plot the entire spectrogram with the audioTestBench function?
classdef FDNreverb2 < audioPlugin
properties
preDelayT = 0; % pre delay [ms]
reverbTime = 1.0; % reverb time [s]
roomSize = 5;
mix = 70; % mix of wet and dry signal [Percent], 100 % -> only wet
order = enumFDNorder.order8; % order of FDN, should be power of 2
in_coeff = 1/2; % coeff for in and output lines --- just for now one value for all
out_coeff = 0.7;
end
properties (Access = private)
N = 8; % order of FDN
a = 1.1; % multiplying factor for delays
cSound = 343.2; % speed of sound
primeDelays = [2 3 5 7 11 13 17 19 23 29 31 37 41 43 47 53 59 61 ...
67 71 73 79 83 89 97 101 103 107 109 113 127 131];
% following values have to be set manually for Interface
maxPreDelay = 0.2; % max pre delay [s] (200ms)
maxRoomSize = 20.0; % max room size [m]
% following variables initialized in reset method
fs; % sample rate
A; % feedback matrix
b; % input gain coefficients
c; % output gain coefficients
f; % feedback lines after matrix
v; % signal before delay - v_i(n) = b_i * x(n) + f_i(n)
s; % output lines
d; % signal to be sent to matrix A
buffDelayLines; % buffer delay lines, initialized in reset method
m; % delay in samples of each delay line
g; % gain coefficients of delay lines
maxDelay; % maximum delay in delay lines
buffInput; % input buffer for pre delay
pBuffDelayLines; % pointer for delay lines buffer
pBuffInput; % pointer for input buffer
preDelayS; % preDelay in samples
alpha;
v_prev2;
v_prev1;
v_filt_prev1;
v_filt_prev2;
end
properties(Constant)
PluginInterface = audioPluginInterface( ...
audioPluginParameter('preDelayT', ...
'DisplayName', 'Pre Delay [ms]', ...
'Mapping', {'int', 0, 200}, ...
'Style', 'rotary', ...
'Layout', [1,1]), ...
audioPluginParameter('roomSize', ...
'DisplayName', 'Room Size [m]', ...
'Mapping', {'lin', 1.0, 20.0}, ...
'Style', 'rotary', ...
'Layout', [1,2]), ...
audioPluginParameter('reverbTime', ...
'DisplayName', 'Reverb Time [s]', ...
'Mapping', {'lin', 0.1, 5.0}, ...
'Style', 'rotary', ...
'Layout', [1,3]), ...
audioPluginParameter('order', ...
'DisplayName', 'Order of FDN', ...
'Mapping', {'enum', '8', '16', '32'}, ...
'Layout', [3,2]), ...
audioPluginParameter('in_coeff', ...
'DisplayName', 'Input Gain', ...
'Mapping', {'lin', 0.0, 1.0}, ...
'Style', 'rotary', ...
'Layout', [3,3]), ...
audioPluginParameter('out_coeff', ...
'DisplayName', 'Output Gain', ...
'Mapping', {'lin', 0.0, 1.0}, ...
'Style', 'rotary', ...
'Layout', [3,4]), ...
audioPluginParameter('mix', ...
'DisplayName', 'Mix', ...
'Mapping', {'int', 0, 100}, ...
'Style', 'rotary', ...
'Layout', [3,1]), ...
audioPluginGridLayout( ...
'RowHeight', [100 100 100 100 100 100], ...
'ColumnWidth',[100, 100 100 100 100 100], ...
'RowSpacing', 10, ...
'ColumnSpacing', 10, ...
'Padding', [10 10 10 10]) ...
);
end
methods
function plugin = FDNreverb
init(plugin)
end
function out = process(plugin, in)
out = zeros(size(in));
%Butterworth coefficients
[bx, ax] = butter(2, 3000/(plugin.fs/2)); % 2. Ordnung, Grenzfrequenz 1000 Hz
b0 = bx(1);
b1 = bx(2);
b2 = bx(3);
a1 = ax(2);
a2 = ax(3);
% rawVn(1) = b0*plugin.v(1);
% rawVn(2) = b0*plugin.v(2) + b1*plugin.v(1) - a1* rawVn(1);
% Define the cutoff frequency and calculate alpha
cutoffFreq = 1000; % Example cutoff frequency in Hz
plugin.alpha = (2 * pi * cutoffFreq) / (plugin.fs + 2 * pi * cutoffFreq);
% Initialize the previous output for the filter
prevY = zeros(plugin.N, 1);
for i = 1:size(in,1)
% Summieren der L/R - Kan�le
inL = in(i,1);
inR = in(i,2);
inSum = (inL + inR)/2;
plugin.buffInput(plugin.pBuffInput + 1) = inSum;
% loop over delay lines
for n=1:plugin.N
% d_n = gain * delayed v_n
for k=1:plugin.N
plugin.d(k) = plugin.g(k) * plugin.buffDelayLines(k, mod(plugin.pBuffDelayLines + plugin.m(k), plugin.maxDelay +1) + 1);
end
% f_n = A(n,:) * d'
plugin.f(n) = plugin.A(n,:) * plugin.d(:);
% v_n with pre delay
rawVn = plugin.b(n) * plugin.buffInput(mod(plugin.pBuffInput + plugin.preDelayS, (plugin.maxPreDelay * plugin.fs + 1)) + 1) + plugin.f(n);
% Filter anwenden
plugin.v(n) = b0 * rawVn + b1 * plugin.v_prev1(n) + b2 * plugin.v_prev2(n) ...
- a1 * plugin.v_filt_prev1(n) - a2 * plugin.v_filt_prev2(n);
% Update der vorherigen Filterwerte
plugin.v_prev2(n) = plugin.v_prev1(n);
plugin.v_prev1(n) = rawVn;
plugin.v_filt_prev2(n) = plugin.v_filt_prev1(n);
plugin.v_filt_prev1(n) = plugin.v(n);
plugin.buffDelayLines(n, plugin.pBuffDelayLines + 1) = plugin.v(n);
% output lines
plugin.s(n) = plugin.c(n) * plugin.d(n);
out(i,:) = out(i,:) + real(plugin.s(n));
end
% Assign to output
out(i,1) = plugin.mix/100 * out(i,1) + (1.0 - plugin.mix/100) * in(i,1);
out(i,2) = plugin.mix/100 * out(i,2) + (1.0 - plugin.mix/100) * in(i,2);
calculatePointer(plugin);
end
end
function calculatePointer(plugin)
if (plugin.pBuffDelayLines==0)
plugin.pBuffDelayLines = plugin.maxDelay;
else
plugin.pBuffDelayLines = plugin.pBuffDelayLines - 1;
end
if (plugin.pBuffInput==0)
plugin.pBuffInput = plugin.maxPreDelay * plugin.fs;
else
plugin.pBuffInput = plugin.pBuffInput - 1;
end
end
function set.in_coeff(plugin, val)
plugin.in_coeff = val;
plugin.b = ones(plugin.N,1) * val;
end
function set.out_coeff(plugin, val)
plugin.out_coeff = val;
plugin.c = ones(plugin.N,1) * val;
end
function set.order(plugin, val)
plugin.order = val;
switch (plugin.order)
case enumFDNorder.order8
plugin.N = 8;
case enumFDNorder.order16
plugin.N = 16;
case enumFDNorder.order32
plugin.N = 32;
end
reset(plugin)
end
function set.reverbTime(plugin, val)
plugin.reverbTime = val;
calculateGainCoeffs(plugin, plugin.reverbTime)
calculateDelays(plugin)
%sprintf('Set Reverb Time: %f', plugin.reverbTime)
%disp(['Set Reverb Time: ', num2str(plugin.reverbTime), ' s.']);
end
function set.preDelayT(plugin, val)
plugin.preDelayT = val;
calculatePreDelayS(plugin, plugin.preDelayT)
%disp(['Set Predelay: ', int2str(plugin.preDelayT), ' ms.']);
end
function set.mix(plugin, val)
plugin.mix = val;
%disp(['Set Mix value: ', int2str(plugin.mix), ' %.']);
end
function calculateMatrix(plugin)
plugin.A = generateFDNmatrix(plugin.N);
end
function calculateDelays(plugin)
% calculate sample delays dependent on room size (val) and
% sample rate (fs)
% m_1 = trace of sound / cSound * fs
M = ceil(0.15 * plugin.reverbTime * plugin.fs);
%disp(['M = ', int2str(M)]);
plugin.m = zeros(plugin.N,1);
interval = M/(plugin.N*4);
%test = 0;
for i=1:plugin.N
tmp = interval/2 + (i-1) * interval;
e = floor(0.5 + log(tmp)/log(plugin.primeDelays(i)));
plugin.m(i) = plugin.primeDelays(i)^(e);
end
end
function calculateGainCoeffs(plugin, val)
for i=1:plugin.N
plugin.g(i) = 10^((-3) * (plugin.m(i)/plugin.fs) / val);
end
end
function calculatePreDelayS(plugin, val)
plugin.preDelayS = val * plugin.fs / 1000;
end
function init(plugin)
plugin.fs = getSampleRate(plugin);
% initialize buffDelayLines & pointer
%plugin.maxDelay = floor(plugin.maxRoomSize/plugin.cSound * plugin.fs * plugin.a^(plugin.N)); %probably higher than actual max delay as actual delays get rounded down
%plugin.maxDelay = ceil(0.15 * 5 * plugin.fs); % maximum possible delay for max reverb time = 5s
plugin.maxDelay = 131^2;
plugin.buffDelayLines = zeros(plugin.N, plugin.maxDelay + 1);
plugin.pBuffDelayLines = plugin.maxDelay;
%initialize filter elements
plugin.v_prev1 = zeros(plugin.N,1);
plugin.v_prev2 = zeros(plugin.N,1);
plugin.v_filt_prev2 = zeros(plugin.N,1);
plugin.v_filt_prev1 = zeros(plugin.N,1);
% initialize
plugin.A = zeros(plugin.N);
plugin.b = ones(plugin.N,1) * plugin.in_coeff; % input gain coefficients
plugin.c = ones(plugin.N,1) * plugin.out_coeff; % output gain coefficients
plugin.f = zeros(plugin.N,1); % feedback lines after matrix
plugin.v = zeros(plugin.N,1); % signal before delay - v_i(n) = b_i * x(n) + f_i(n)
plugin.s = zeros(plugin.N,1); % output lines
plugin.d = zeros(plugin.N,1); % signal to be sent fsto matrix A
plugin.m = zeros(plugin.N,1); % delay in samples of each delay line
plugin.g = zeros(plugin.N,1); % gain coefficients of delay lines
plugin.buffInput = zeros((plugin.maxPreDelay * plugin.fs) + 1, 1); % input buffer for pre delay, 0.2 = max pre delay of 200ms
plugin.preDelayS = 0; % pre delay in samples
plugin.pBuffInput = plugin.maxPreDelay * plugin.fs; % pointer for input buffer
% calculate sample delays of delay lines
calculateDelays(plugin)
% calculate gain coeffs of delay lines
calculateGainCoeffs(plugin, plugin.reverbTime)
% calculate FDN matrix based on set order
calculateMatrix(plugin)
end
function reset(plugin)
init(plugin)
end
function generateFDNmatrix(order)
% generates FDN matrix with evenly distributed eigenvalues at
% the unit circle
eig_nr = order/2;
FDNmatrix = zeros(order, order);
M = orth((rand(order,order)));
BC_n = zeros(order, order, eig_nr);
DtD_n = zeros(order, order, eig_nr);
k1 = 1;
for k=2:2:eig_nr*2
x = M(:,k-1);
y = M(:,k);
x = x / sqrt(2);
y = y / sqrt(2);
B = x * x.';
C = y * y.';
D = x * y.';
BC_n(1:order,1:order,k1) = B + C;
DtD_n(1:order,1:order,k1) = D.' - D;
w = pi * k / (eig_nr * 2 + 2);
temp = 2 * BC_n(:,:,k1) * cos(w) + 2 * DtD_n(:,:,k1) * sin(w);
FDNmatrix = FDNmatrix + temp;
k1 = k1 + 1;
end
end
end
end

Answers (2)

Ayush
Ayush on 17 Sep 2024
Edited: Ayush on 17 Sep 2024
I understand that you need to generate and plot the entire spectrogram of signal processed by the audio plugin. To achieve this, you need to capture the output of the plugin over time and then use functions like “spectrogram” in MATLAB to visualize it.
The “audioTestBench” function in MATLAB is primarily for testing and tuning audio plugins interactively, but it doesn’t provide a way to capture the entire output for spectrogram plotting.
Here’s a code snippet to illustrate the above approach:
% Load the audio signal
[inputSignal, fs] = audioread('your_audio_file.wav'); % your_audio_file : sample audio file
% Creating the plugin object
plugin = FDNreverb2; %FDNreverb2 is plugin class as in your code
% Process the input signal
outputSignal = plugin.process(inputSignal);
% Compute and plot the spectrogram
windowLength = 1024;
overlap = 512;
nfft = 2048;
spectrogram(outputSignal(:,1), windowLength, overlap, nfft, fs, 'yaxis');
title('Spectrogram of Processed Signal');
xlabel('Time (s)');
ylabel('Frequency (Hz)');
I've taken a sample audio file and following is the spectrogram I obtained:
You can read more about “audioread” function here:
You can read more about “spectrogram” here:
Hope this helps!

jibrahim
jibrahim on 30 Oct 2024
audioTestBench has a spectrum analyzer which includes a spectrogram option.
Launch the spectrum analyzer from audioTstBench by clicking the Spectrum analyzer button in the toopstrip.
In the spectrum analyzer, turn on spectrogram mode by going under the scope menu and clicking spectrogram.
Now, when you execute audioTestBench, you will able to see the spectrogram of the output signal as the algorithm is running.

Categories

Find more on Audio Processing Algorithm Design in Help Center and File Exchange

Products


Release

R2024a

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!