The Auditory Modeling Toolbox
Go to function
EXP_LOPEZPOVEDA2001 - Figures from Lopez-Poveda and Meddis (2001)
Program code:
function output=exp_lopezpoveda2001(varargin)
%EXP_LOPEZPOVEDA2001 Figures from Lopez-Poveda and Meddis (2001)
% Usage: output = exp_lopezpoveda2001(flag)
%
% EXP_LOPEZPOVEDA2001(flags,... ) reproduces experiments from the Lopez
% & Poveda (2001) paper.
%
% The following flags can be specified;
%
% 'plot' Plot the output of the experiment. This is the default.
%
% 'noplot' Don't plot, only return data.
%
% 'fig2' Reproduce Fig. 2 from Lopez & Poveda (2001)
%
% Fig. 2a represents the outer ear filter - pressure gain (dB) over
% frequency with data points from Pralong and Carlile (1996).
%
% Fig. 2b represents the middle ear filter - stapes velocity at
% 0dB over frequency in one plot Fig. 2b shows for default
% fs = 22050 Hz:
%
% Data points directly derived from Goode et al. 1994
%
% FIR filter with data points from Goode et al. 1994
%
% Data points of figure 2b from Lopez-Poveda and Meddis
% 2001 (read from fig 2b, actually also derived from Goode et
% al. 1994) The output are the data points of the respective
% figure.
%
% The dimensions of the output are: frequency values x data
% points x figure no.
%
% 'fig2a' Reproduce just Fig. 2a.
%
% 'fig2b' Reproduce just Fig. 2b.
%
% 'fig3bc' Reproduce Fig. 3b and c from Lopez & Poveda
% (2001). Isointensity response of the linear, nonlinear and
% summed response of the DRNL filter for an input level of
% 30dB (fig 3b) and 85dB (fig 3c) SPL The output is the output
% of the DRNL filter for the different input levels. The
% dimensions of the output are: input frequency x [frequency
% values, linear output, nonlinear output, summed DRNL output]
% x input level.
%
% 'fig3b' Reproduce just Fig. 3b.
%
% 'fig3c' Reproduce just fig. 3c.
%
% 'fig4' Reproduce Fig. 4 from Lopez & Poveda (2001) - pulsation threshold
% data (just the model results, not the experimental data)
% The output is the model result for the different parameter sets.
% The dimensions of the output are: signal level x masker level x signal frequency
% masker level consists of 4 columns:
%
% 1) Signal level in dB SPL (x axis in the plots)
%
% 2) Results for parameter set of YO, table I
%
% 3) Results for average parameter set, table II
%
% 4) Results for regression lines, table III
%
% See also: drnl, data_lopezpoveda2001, data_pralong1996, data_goode1994
%
% Examples:
% ---------
%
% To display Figure 2 use :
%
% exp_lopezpoveda2001('fig2');
%
% To display Figure 3b and 3c use :
%
% exp_lopezpoveda2001('fig3bc');
%
% To display Figure 4 use :
%
% exp_lopezpoveda2001('fig4');
%
% References:
% R. Goode, M. Killion, K. Nakamura, and S. Nishihara. New knowledge
% about the function of the human middle ear: development of an improved
% analog model. The American journal of otology, 15(2):145-154, 1994.
%
% E. Lopez-Poveda and R. Meddis. A human nonlinear cochlear filterbank.
% J. Acoust. Soc. Am., 110:3107-3118, 2001.
%
% D. Pralong and S. Carlile. The role of individualized headphone
% calibration for the generation of high fidelity virtual auditory space.
% J. Acoust. Soc. Am., 100:3785-3793, 1996.
%
%
% Url: http://amtoolbox.sourceforge.net/data/amt-test/htdocs/amt-0.9.8/doc/experiments/exp_lopezpoveda2001.php
% Copyright (C) 2009-2015 Piotr Majdak and Peter L. Søndergaard.
% This file is part of AMToolbox version 0.9.8
%
% This program is free software: you can redistribute it and/or modify
% it under the terms of the GNU General Public License as published by
% the Free Software Foundation, either version 3 of the License, or
% (at your option) any later version.
%
% This program is distributed in the hope that it will be useful,
% but WITHOUT ANY WARRANTY; without even the implied warranty of
% MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
% GNU General Public License for more details.
%
% You should have received a copy of the GNU General Public License
% along with this program. If not, see <http://www.gnu.org/licenses/>.
% AUTHOR: Katharina Egger
%% ------ Check input options --------------------------------------------
definput.flags.type = {'missingflag','fig2','fig2a','fig2b','fig3bc','fig3b','fig3c','fig4'};
definput.flags.plot = {'plot','noplot'};
definput.keyvals.predrnl = {};
definput.keyvals.postdrnl = {};
% Parse input options
[flags,kv] = ltfatarghelper({},definput,varargin);
if flags.do_missingflag
flagnames=[sprintf('%s, ',definput.flags.type{2:end-2}),...
sprintf('%s or %s',definput.flags.type{end-1},definput.flags.type{end})];
error('%s: You must specify one of the following flags: %s.',upper(mfilename),flagnames);
end;
%% parameter set of YO, table I
% The data is specified in this way, because the data for figure 4 is
% not specified as the two parameter fit, but instead specified
% directly. The parameters below accomplish this by removing all
% frequency dependence in the DRNL. The parameters can then be
% specified exactly for a single frequency, but only for one frequency
% at a time.
f250 = {...
'flow',250,...
'fhigh',250,...
'lin_fc', [log10(235) 0],...
'lin_bw', [log10(115) 0],...
'lin_gain', [log10(1400) 0],...
'lin_lp_cutoff', [log10(235) 0],...
'nlin_fc_before', [log10(250) 0],...
'nlin_bw_before', [log10(84) 0],...
'nlin_lp_cutoff', [log10(250) 0],...
'nlin_a', [log10(2124) 0],...
'nlin_b', [log10(0.45) 0] };
f500 = {...
'flow',500,...
'fhigh',500,...
'lin_fc', [log10(460) 0],...
'lin_bw', [log10(150) 0],...
'lin_gain', [log10(800) 0],...
'lin_lp_cutoff', [log10(460) 0],...
'nlin_fc_before', [log10(500) 0],...
'nlin_bw_before', [log10(103) 0],...
'nlin_lp_cutoff', [log10(500) 0],...
'nlin_a', [log10(4609) 0],...
'nlin_b', [log10(0.28) 0] };
f1000 = {...
'flow',1000,...
'fhigh',1000,...
'lin_fc', [log10(945) 0],...
'lin_bw', [log10(240) 0],...
'lin_gain', [log10(520) 0],...
'lin_lp_cutoff', [log10(945) 0],...
'nlin_fc_before', [log10(1000) 0],...
'nlin_bw_before', [log10(175) 0],...
'nlin_lp_cutoff', [log10(1000) 0],...
'nlin_a', [log10(4598) 0],...
'nlin_b', [log10(0.130) 0]};
f2000 = {...
'flow',2000,...
'fhigh',2000,...
'lin_fc', [log10(1895) 0],...
'lin_bw', [log10(390) 0],...
'lin_gain', [log10(400) 0],...
'lin_lp_cutoff', [log10(1895) 0],...
'nlin_fc_before', [log10(2000) 0],...
'nlin_bw_before', [log10(300) 0],...
'nlin_lp_cutoff', [log10(2000) 0],...
'nlin_a', [log10(9244) 0],...
'nlin_b', [log10(0.078) 0]};
f4000 = {...
'flow',4000,...
'fhigh',4000,...
'lin_fc', [log10(3900) 0],...
'lin_bw', [log10(620) 0],...
'lin_gain', [log10(270) 0],...
'lin_lp_cutoff', [log10(3900) 0],...
'nlin_fc_before', [log10(4000) 0],...
'nlin_bw_before', [log10(560) 0],...
'nlin_lp_cutoff', [log10(4000) 0],...
'nlin_a', [log10(30274) 0],...
'nlin_b', [log10(0.06) 0]};
f8000 = {...
'flow',8000,...
'fhigh',8000,...
'lin_fc', [log10(7450) 0],...
'lin_bw', [log10(1550) 0],...
'lin_gain', [log10(250) 0],...
'lin_lp_cutoff', [log10(7450) 0],...
'nlin_fc_before', [log10(8000) 0],...
'nlin_bw_before', [log10(1100) 0],...
'nlin_lp_cutoff', [log10(8000) 0],...
'nlin_a', [log10(76354) 0],...
'nlin_b', [log10(0.035) 0]};
expparsYO = [f250; f500; f1000; f2000; f4000; f8000];
%% average parameter set, table II
f250avg = {...
'flow',250,...
'fhigh',250,...
'lin_fc', [log10(244) 0],...
'lin_bw', [log10(100) 0],...
'lin_gain', [log10(1150) 0],...
'lin_lp_cutoff', [log10(244) 0],...
'lin_ngt', 3,...
'nlin_fc_before', [log10(250) 0],...
'nlin_bw_before', [log10(84) 0],...
'nlin_lp_cutoff', [log10(250) 0],...
'nlin_a', [log10(2194) 0],...
'nlin_b', [log10(0.45) 0]};
f500avg = {...
'flow',500,...
'fhigh',500,...
'lin_fc', [log10(480) 0],...
'lin_bw', [log10(130) 0],...
'lin_gain', [log10(850) 0],...
'lin_lp_cutoff', [log10(480) 0],...
'lin_ngt', 3,...
'nlin_fc_before', [log10(500) 0],...
'nlin_bw_before', [log10(103) 0],...
'nlin_lp_cutoff', [log10(500) 0],...
'nlin_a', [log10(5184) 0],...
'nlin_b', [log10(0.28) 0]};
f1000avg = {...
'flow',1000,...
'fhigh',1000,...
'lin_fc', [log10(965) 0],...
'lin_bw', [log10(240) 0],...
'lin_gain', [log10(520) 0],...
'lin_lp_cutoff', [log10(965) 0],...
'lin_ngt', 3,...
'nlin_fc_before', [log10(1000) 0],...
'nlin_bw_before', [log10(175) 0],...
'nlin_lp_cutoff', [log10(1000) 0],...
'nlin_a', [log10(7558) 0],...
'nlin_b', [log10(0.130) 0]};
f2000avg = {...
'flow',2000,...
'fhigh',2000,...
'lin_fc', [log10(1925) 0],...
'lin_bw', [log10(400) 0],...
'lin_gain', [log10(410) 0],...
'lin_lp_cutoff', [log10(1925) 0],...
'lin_ngt', 3,...
'nlin_fc_before', [log10(2000) 0],...
'nlin_bw_before', [log10(300) 0],...
'nlin_lp_cutoff', [log10(2000) 0],...
'nlin_a', [log10(9627) 0],...
'nlin_b', [log10(0.078) 0]};
f4000avg = {...
'flow',4000,...
'fhigh',4000,...
'lin_fc', [log10(3900) 0],...
'lin_bw', [log10(660) 0],...
'lin_gain', [log10(320) 0],...
'lin_lp_cutoff', [log10(3900) 0],...
'lin_ngt', 3,...
'nlin_fc_before', [log10(4000) 0],...
'nlin_bw_before', [log10(560) 0],...
'nlin_lp_cutoff', [log10(4000) 0],...
'nlin_a', [log10(22288) 0],...
'nlin_b', [log10(0.045) 0]};
f8000avg = {...
'flow',8000,...
'fhigh',8000,...
'lin_fc', [log10(7750) 0],...
'lin_bw', [log10(1450) 0],...
'lin_gain', [log10(220) 0],...
'lin_lp_cutoff', [log10(7750) 0],...
'lin_ngt', 3,...
'nlin_fc_before', [log10(8000) 0],...
'nlin_bw_before', [log10(1100) 0],...
'nlin_lp_cutoff', [log10(8000) 0],...
'nlin_a', [log10(43584) 0],...
'nlin_b', [log10(0.030) 0]};
expparsavg = [f250avg; f500avg; f1000avg; f2000avg; f4000avg; f8000avg];
%% Lopez-Poveda and Meddis 2001, Figure 2
%% Lopez-Poveda and Meddis 2001, Figure 2, a)
if flags.do_fig2a || flags.do_fig2
fs=22050;
hpdata=data_pralong1996;
bout=headphonefilter(fs);
% Manually calculate the frequency response.
fout = 20*log10(abs(fftreal(bout)));
% Half the filter length.
n2=length(fout);
output(:,:,1) = hpdata;
end
%% Lopez-Poveda and Meddis 2001, Figure 2, b)
if flags.do_fig2b || flags.do_fig2
fs = 22050;
% data points directly derived from Goode et al. 1994
gde = middleearfilter;
% control data points directly read from Lopez-Poveda and Meddis 2001
stapes_data = data_lopezpoveda2001('fig2b', 'noplot', 'lopezpoveda');
% Calculate the filters.
bmid = middleearfilter(fs); % Goode et al. 1994
% Manually calculate the frequency response for an input of 0dB SPL
fmid = abs(fftreal(bmid*20e-6));
% Half the filter length.
n2=length(fmid);
% x-values for plotting.
xplot=linspace(0,fs/2,n2);
outB = gde;
if exist('output','var')
output(end+1:end+(length(outB)-length(output)),:,1) = 0;
output(:,:,2) = outB;
else
output(:,:,1) = outB;
end
end
%% plots
if flags.do_plot
if flags.do_fig2
figure
set(gcf,'Position',[50,50,500,760])
subplot(2,1,1)
hold on;
% Plot the measured data
x=hpdata(:,1);
freqresp=20*log10(hpdata(:,2));
plot(x,freqresp,'ro','MarkerFaceColor', 'r');
% Plot the filter
x_filter=linspace(0,fs/2,n2);
plot(x_filter,fout);
axis([100 10000 -30 20])
set(gca,'XScale','log','YTick',[-30,-20,-10,0,10,20])
set(gca,'Position',[0.15,0.55,0.8,0.4]);
leg1=legend('Pralong and Carlile (1996) + extrapolated points', ...
'FIR filter');
xlabel('Frequency (Hz)')
ylabel('Pressure gain (dB)')
title('Lopez-Poveda and Meddis 2001, Figure 2')
set(leg1,'Position',[0.1887, 0.5991, 0.666, 0.0553]);
subplot(2,1,2)
p = loglog (stapes_data(:,1),stapes_data(:,2),':ok', 'MarkerFaceColor', 'k');
hold on
g = loglog (gde(:,1),gde(:,2),':or', 'MarkerFaceColor', 'r');
firG = loglog(xplot,fmid,'r');
axis([100 10000 1E-10 1E-07])
set(gca,'Position',[0.15,0.07,0.8,0.4]);
xlabel('Frequency (Hz)')
ylabel('Stapes velocity (m/s) at 0dB SPL')
leg2=legend([g,firG,p],'directly derived from Goode et al. 1994', ...
'FIR filter with data points from Goode et al. 1994', ...
'Control points as in Lopez-Poveda and Meddis 2001');
set(leg2,'Position',[0.1637, 0.085, 0.716, 0.07]);
elseif flags.do_fig2a
hold on;
% Plot the measured data
x=hpdata(:,1);
freqresp=20*log10(hpdata(:,2));
plot(x,freqresp,'ro','MarkerFaceColor', 'r');
% Plot the filter
x_filter=linspace(0,fs/2,n2);
plot(x_filter,fout);
axis([100 10000 -30 20])
set(gca,'XScale','log','YTick',[-30,-20,-10,0,10,20])
legend('Pralong and Carlile (1996) + extrapolated points', ...
'FIR filter');
title('Lopez-Poveda and Meddis 2001, Figure 2a) - Pressure gain (dB) as a function of frequency')
xlabel('Frequency (Hz)')
ylabel('Pressure gain (dB)')
hold off
elseif flags.do_fig2b
p = loglog (stapes_data(:,1),stapes_data(:,2),':ok', 'MarkerFaceColor', 'k');
hold on
g = loglog (gde(:,1),gde(:,2),':or', 'MarkerFaceColor', 'r');
firG = loglog(xplot,fmid,'r');
axis([100 10000 1E-10 1E-07])
title('Lopez-Poveda and Meddis 2001, Figure 2b) - Stapes velocity as a function of frequency')
xlabel('Frequency (Hz)')
ylabel('Stapes velocity (m/s) at 0dB SPL')
legend([g,firG,p],'directly derived from Goode et al. 1994', 'FIR filter with data points from Goode et al. 1994', ...
'Control points as in Lopez-Poveda and Meddis 2001')
hold off
end
end
%% Lopez-Poveda and Meddis 2001, Figure 3
if flags.do_fig3b || flags.do_fig3bc
%% Lopez-Poveda and Meddis 2001, Figure 3, b)
% input signal: 50ms pure tones, sampled at 100kHz
fs = 100000;
T = 0.05;
t = (0:1/fs:T - 1/fs)';
fsig = 250:25:1750;
result3b = zeros(1,length(fsig));
lin3b = zeros(1,length(fsig));
nlin3b = zeros(1,length(fsig));
level = 20e-6 * 10^(30/20);
for ii = 1:length(fsig)
insig = sin(2*pi*fsig(ii).*t)*(2^0.5) * level;
hp_fir = headphonefilter(fs);
insig = filter(hp_fir,1,insig);
[y_lin, ~] = drnl(insig, fs, kv.predrnl{:}, f1000{:},'linonly', kv.postdrnl{:});
[y_nlin, ~] = drnl(insig, fs, kv.predrnl{:},f1000{:},'nlinonly', kv.postdrnl{:});
outsig = y_lin + y_nlin;
result3b(1,ii) = rms(outsig(floor(length(insig)/2):end));
lin3b(1,ii) = rms(y_lin(floor(length(insig)/2):end));
nlin3b(1,ii) = rms(y_nlin(floor(length(insig)/2):end));
end
output(:,:,1) = [fsig', lin3b', nlin3b', result3b'];
end;
if flags.do_fig3c || flags.do_fig3bc
%% Lopez-Poveda and Meddis 2001, Figure 3, c)
% input signal: 50ms pure tones, sampled at 100kHz
fs = 100000;
T = 0.05;
t = (0:1/fs:T - 1/fs)';
fsig = 250:25:1750;
result3c = zeros(1,length(fsig));
lin3c = zeros(1,length(fsig));
nlin3c = zeros(1,length(fsig));
level = 20e-6 * 10^(85/20);
for ii = 1:length(fsig)
insig = sin(2*pi*fsig(ii).*t)*(2^0.5)* level;
hp_fir = headphonefilter(fs);
insig = filter(hp_fir,1,insig);
[y_lin, ~] = drnl(insig, fs, kv.predrnl{:}, f1000{:},'linonly', kv.postdrnl{:});
[y_nlin, ~] = drnl(insig, fs, kv.predrnl{:}, f1000{:},'nlinonly', kv.postdrnl{:});
outsig = y_lin + y_nlin;
result3c(1,ii) = rms(outsig(floor(length(insig)/2):end));
lin3c(1,ii) = rms(y_lin(floor(length(insig)/2):end));
nlin3c(1,ii) = rms(y_nlin(floor(length(insig)/2):end));
end
outB = [fsig', lin3c', nlin3c', result3c'];
if exist('output','var')
output(:,:,2) = outB;
else
output(:,:,1) = outB;
end
end;
%% plots
if flags.do_plot
if flags.do_fig3bc
figure
set(gcf,'Position',[50,50,400,760])
subplot(2,1,1)
plot(fsig,result3b)
hold on
plot(fsig,lin3b,'-.g')
plot(fsig,nlin3b,':r')
set(gca,'YScale','log')
% grid on
set(gca,'XLim',[250 1750],'Layer','top')
set(gca,'YLim',[1e-07 1e-03],'Layer','top')
set(gca,'Position',[0.285,0.5838,0.62,0.3412]);
title('30 dB SPL')
xlabel('Frequency (Hz)')
ylabel('DRNL filter output (m/s)')
subplot(2,1,2)
plot(fsig,result3c)
hold on
plot(fsig,lin3c,'-.g')
plot(fsig,nlin3c,':r')
set(gca,'YScale','log')
% grid on
set(gca,'XLim',[250 1750],'Layer','top')
set(gca,'YLim',[1e-05 1e-01],'Layer','top')
set(gca,'Position',[0.285,0.11,0.62,0.3412]);
title('85 dB SPL')
xlabel('Frequency (Hz)')
ylabel('DRNL filter output (m/s)')
leg=legend('DRNL output', 'Linear path output', 'Nonlinear path output');
set(leg,'Position',[0.0133, 0.4759, 0.4525, 0.0798]);
elseif flags.do_fig3b
plot(fsig,result3b)
hold on
plot(fsig,lin3b,'-.g')
plot(fsig,nlin3b,':r')
set(gca,'YScale','log')
set(gca,'XLim',[250 1750],'Layer','top')
set(gca,'YLim',[1e-07 1e-03],'Layer','top')
title('30 dB SPL')
xlabel('Frequency (Hz)')
ylabel('DRNL filter output (m/s)')
leg=legend('DRNL output', 'Linear path output', 'Nonlinear path output');
set(gcf,'Position',[150,150,400,400])
set(leg,'Position',[0.2333, 0.1467, 0.4525, 0.1517]);
hold off
elseif flags.do_fig3c
plot(fsig,result3c)
hold on
plot(fsig,lin3c,'-.g')
plot(fsig,nlin3c,':r')
set(gca,'YScale','log')
set(gca,'XLim',[250 1750],'Layer','top')
set(gca,'YLim',[1e-05 1e-01],'Layer','top')
title('85 dB SPL')
xlabel('Frequency (Hz)')
ylabel('DRNL filter output (m/s)')
leg=legend('DRNL output', 'Linear path output', 'Nonlinear path output');
set(gcf,'Position',[150,150,400,400])
set(leg,'Position',[0.2333, 0.1467, 0.4525, 0.1517]);
hold off
end
end
%% Lopez-Poveda and Meddis 2001, Figure 4
if flags.do_fig4
%% input signal
fs=64000;
fsig = [250 500 1000 2000 4000 8000];
T = 0.084;
t = (0:1/fs:T - 1/fs)';
%basef = fsig;
hp_fir = headphonefilter(fs);
Tramp = 0.002; % duration of ramps up and down
ramp = (0:1/fs:Tramp - 1/fs)';
sig=zeros(length(t),length(fsig));
mask=zeros(length(t),length(fsig));
%% experiment
LSDB = 30:0.5:85; % Signal level
n=1;
for jj = 30:0.5:85
levelS(n) = 20e-6 * 10^(jj/20);
n = n+1;
end
LMDB = 30:0.5:100; % Masker level
n=1;
for jj = 30:0.5:100
levelM(n) = 20e-6 * 10^(jj/20);
n = n+1;
end
OMavg = zeros(length(levelM),length(fsig));
OM = zeros(length(levelM),length(fsig));
OSavg = zeros(length(levelS),length(fsig));
OS = zeros(length(levelS),length(fsig));
ratio = zeros(length(levelM),length(levelS),length(fsig));
ratioavg = zeros(length(levelM),length(levelS),length(fsig));
indx = zeros(length(levelS),length(fsig));
indxavg = zeros(length(levelS),length(fsig));
output = zeros(length(LSDB),4,length(fsig));
for ii = 1:length(fsig)
% first calculate the model's response to the masker for every
% possible masker level
for kk = 1:length(levelM)
% rampsignal sine window equivalent to cosine ramps are used
mask(:,ii) = rampsignal(sin(2*pi*fsig(ii)*0.6.*t),length(ramp),'sine').*(2^0.5);
insig = mask(:,ii) * levelM(kk);
outsig = filter(hp_fir,1,insig);
% average parameter set, table II
outsigavg = drnl(outsig, fs, kv.predrnl{:}, expparsavg{ii,:}, kv.postdrnl{:});
OMavg(kk,ii) = max(outsigavg(floor(length(insig)/2):end));
% parameter set of YO, table I
outsig = drnl(outsig, fs, kv.predrnl{:}, expparsYO{ii,:}, kv.postdrnl{:});
OM(kk,ii) = max(outsig(floor(length(insig)/2):end));
end
% then calculate model's response to the signal and find for every
% signal level the masker level such that the ratio signal/masker
% is equal to a value of one
for mm = 1:length(levelS)
sig(:,ii) = rampsignal(sin(2*pi*fsig(ii).*t),length(ramp),'sine').*(2^0.5);
insig = sig(:,ii) * levelS(mm);
outsig = filter(hp_fir,1,insig);
% average parameter set, table II
outsigavg = drnl(outsig, fs, kv.predrnl{:}, expparsavg{ii,:}, kv.postdrnl{:});
OSavg(mm,ii) = max(outsigavg(floor(length(insig)/2):end));
% parameter set of YO, table I
outsig = drnl(outsig, fs, kv.predrnl{:}, expparsYO{ii,:}, kv.postdrnl{:});
OS(mm,ii) = max(outsig(floor(length(insig)/2):end));
% ratio for parameter set of YO, table I
ratio(:,mm,ii) = OS(mm,ii) ./ OM(:,ii);
[~, indx(mm,ii)] = min(abs(1-ratio(:,mm,ii)),[],1);
% ratio for average parameter set, table II
ratioavg(:,mm,ii) = OSavg(mm,ii) ./ OMavg(:,ii);
[~, indxavg(mm,ii)] = min(abs(1-ratioavg(:,mm,ii)),[],1);
end
output(:,1,ii) = LSDB;
output(:,2,ii) = LMDB(indx(:,ii));
output(:,3,ii) = LMDB(indxavg(:,ii));
end
% same procedure for DRNL parameters calculated with regression lines, table III
OMrl = zeros(length(levelM),length(fsig));
OSrl = zeros(length(levelS),length(fsig));
ratiorl = zeros(length(levelM),length(levelS),length(fsig));
indxrl = zeros(length(levelS),length(fsig));
for ii = 1:length(fsig)
for kk = 1:length(levelM)
insig = mask(:,ii) * levelM(kk);
outsig = filter(hp_fir,1,insig);
outsigrl = drnl(outsig, fs, kv.predrnl{:}, 'flow', fsig(ii), 'fhigh', fsig(ii), 'lin_ngt', 3, kv.postdrnl{:});
OMrl(kk,ii) = max(outsigrl(floor(length(insig)/2):end));
end
for mm = 1:length(levelS)
insig = sig(:,ii) * levelS(mm);
outsig = filter(hp_fir,1,insig);
outsigrl = drnl(outsig, fs, kv.predrnl{:}, 'flow', fsig(ii), 'fhigh', fsig(ii), 'lin_ngt', 3, kv.postdrnl{:});
OSrl(mm,ii) = max(outsigrl(floor(length(insig)/2):end));
ratiorl(:,mm,ii) = OSrl(mm,ii) ./ OMrl(:,ii);
[~, indxrl(mm,ii)] = min(abs(1-ratiorl(:,mm,ii)),[],1);
end
output(:,4,ii) = LMDB(indxrl(:,ii));
end
%% plots
if flags.do_plot
subplot(2,3,1)
plot(LSDB,LMDB(indx(:,1)),'-', 'LineWidth', 2)
hold on
plot(LSDB,LMDB(indxavg(:,1)),'-')
plot(LSDB,LMDB(indxrl(:,1)),'--')
plot(LSDB,LSDB,':')
grid on
set(gca,'XLim',[20 90],'Layer','top','YTick',[55,65,75,85,95])
axis([20,90,55,100]);
title('250 Hz')
xlabel('Signal level (dB SPL)')
ylabel('Masker level (dB SPL)')
subplot(2,3,2)
plot(LSDB,LMDB(indx(:,2)),'-', 'LineWidth', 2)
hold on
plot(LSDB,LMDB(indxavg(:,2)),'-')
plot(LSDB,LMDB(indxrl(:,2)),'--')
plot(LSDB,LSDB,':')
grid on
set(gca,'XLim',[20 90],'Layer','top','YTick',[55,65,75,85,95])
axis([20,90,55,100]);
title('500 Hz')
xlabel('Signal level (dB SPL)')
ylabel('Masker level (dB SPL)')
subplot(2,3,3)
plot(LSDB,LMDB(indx(:,3)),'-', 'LineWidth', 2)
hold on
plot(LSDB,LMDB(indxavg(:,3)),'-')
plot(LSDB,LMDB(indxrl(:,3)),'--')
plot(LSDB,LSDB,':')
grid on
set(gca,'XLim',[20 90],'Layer','top','YTick',[55,65,75,85,95])
axis([20,90,55,100]);
title('1000 Hz')
xlabel('Signal level (dB SPL)')
ylabel('Masker level (dB SPL)')
subplot(2,3,4)
plot(LSDB,LMDB(indx(:,4)),'-', 'LineWidth', 2)
hold on
plot(LSDB,LMDB(indxavg(:,4)),'-')
plot(LSDB,LMDB(indxrl(:,4)),'--')
plot(LSDB,LSDB,':')
grid on
set(gca,'XLim',[20 90],'Layer','top','YTick',[55,65,75,85,95])
axis([20,90,55,100]);
title('2000 Hz')
xlabel('Signal level (dB SPL)')
ylabel('Masker level (dB SPL)')
subplot(2,3,5)
plot(LSDB,LMDB(indx(:,5)),'-', 'LineWidth', 2)
hold on
plot(LSDB,LMDB(indxavg(:,5)),'-')
plot(LSDB,LMDB(indxrl(:,5)),'--')
plot(LSDB,LSDB,':')
grid on
set(gca,'XLim',[20 90],'Layer','top','YTick',[55,65,75,85,95])
axis([20,90,55,100]);
title('4000 Hz')
xlabel('Signal level (dB SPL)')
ylabel('Masker level (dB SPL)')
subplot(2,3,6)
plot(LSDB,LMDB(indx(:,6)),'-', 'LineWidth', 2)
hold on
plot(LSDB,LMDB(indxavg(:,6)),'-')
plot(LSDB,LMDB(indxrl(:,6)),'--')
plot(LSDB,LSDB,':')
grid on
set(gca,'XLim',[20 90],'Layer','top','YTick',[55,65,75,85,95])
axis([20,90,55,100]);
title('8000 Hz')
xlabel('Signal level (dB SPL)')
ylabel('Masker level (dB SPL)')
legend('parameter set of YO, table I','average parameter set, table II','regression lines, table III','linear behavior')
end
end;
Build with Bootstrap