THE AUDITORY MODELING TOOLBOX
Go to function
EXP_ENGEL2022
Effects of binaural-rendering techniques assessed by various models
Program code:
function exp_engel2022(varargin)
%EXP_ENGEL2022 Effects of binaural-rendering techniques assessed by various models
%
% Usage: data = exp_engel2022(flag)
%
% EXP_ENGEL2022(flag) reproduces figures of the study from
% Engel et al. (2022). This article investigates the effects of various
% HRTF preprocessing methods for Ambisonics rendering. The effects are
% assessed by applying a model for sound localization REIJNIERS2014,
% a model for sound externalization BAUMGARTNER2021, and a model for
% speech reception JELFS2011.
%
% The following flags can be specified:
%
% 'fig1' Spherical harmonics spectra
%
% 'fig2' Mag/phase errors per spatial order
%
% 'fig3' ITD/ILD per spatial order
%
% 'fig4' Mag/phase errors per method
%
% 'fig5' ITD/ILD per method
%
% 'fig6' Loudness per direction
%
% 'fig7' Models' outputs per spatial order
%
%
% For the calculations, an HRTF data set will be loaded from auxdata and
% represented as truncated spherical-harmonics HRTFs. Then, various
% rendering techniques will be applied and the results will be provided
% as input to the models.
%
% Warning: The processing of HRTFs does not take a long time to generate
% (minutes). The model results may take long time to generate (hours).
% Both, the processed HRTFs and model results are available as cached data.
%
%
% Examples:
% ---------
%
% To display Fig.1 use :
%
% exp_engel2022('fig1');
%
% To display Fig.2 use :
%
% exp_engel2022('fig2');
%
% To display Fig.3 use :
%
% exp_engel2022('fig3');
%
% To display Fig.4 use :
%
% exp_engel2022('fig4');
%
% To display Fig.5 use :
%
% exp_engel2022('fig5');
%
% To display Fig.6 use :
%
% exp_engel2022('fig6');
%
% To display Fig.7 use :
%
% exp_engel2022('fig7');
%
%
% References:
% Engel, Isaac, Goodman, Dan F. M., and Picinali, Lorenzo. Assessing hrtf
% preprocessing methods for ambisonics rendering through perceptual
% models. Acta Acust., 6:4, 2022.
%
% R. Baumgartner and P. Majdak. Decision making in auditory
% externalization perception: model predictions for static conditions.
% Acta Acustica, 5:59, 2021. Publisher: EDP Sciences.
%
% S. Jelfs, J. Culling, and M. Lavandier. Revision and validation of a
% binaural model for speech intelligibility in noise. Hearing Research,
% 2011.
%
% J. Reijniers, D. Vanderleist, C. Jin, C. S., and H. Peremans. An
% ideal-observer model of human sound localization. Biological
% Cybernetics, 108:169--181, 2014.
%
%
% See also: baumgartner2021 jelfs2011 reijniers2014
%
% Url: http://amtoolbox.org/amt-1.6.0/doc/experiments/exp_engel2022.php
% #Requirements: BinauralSH
% #Author: Isaac Engel (2021): initial implementation
% #Author: Piotr Majdak (2023): adaptation for the AMT 1.5.0
% This file is licensed unter the GNU General Public License (GPL) either
% version 3 of the license, or any later version as published by the Free Software
% Foundation. Details of the GPLv3 can be found in the AMT directory "licences" and
% at <https://www.gnu.org/licenses/gpl-3.0.html>.
% You can redistribute this file and/or modify it under the terms of the GPLv3.
% This file is distributed without any warranty; without even the implied warranty
% of merchantability or fitness for a particular purpose.
definput.flags.experiment = {'missingflag',...
'fig1','fig2','fig3','fig4','fig5','fig6','fig7'};
definput.import={'amt_cache'};
[flags,~] = ltfatarghelper({},definput,varargin);
if flags.do_missingflag
flagnames=[sprintf('%s, ',definput.flags.experiment{2:end-2}),...
sprintf('%s or %s',definput.flags.experiment{end-1},definput.flags.experiment{end})];
error('%s: You must specify one of the following flags: %s.',upper(mfilename),flagnames);
end
%% Parameters
N_vec = 1:44; % spatial orders to be tested
Nref = 44; % reference (very high) spatial order
r = 0.0875; % nominal head radius = 8.75 cm
itdJND = 20; % ITD JND for plots, according to Klockgether 2016
ildJND = 0.6; % ILD JND for plots, according to Klockgether 2016
pagewidth = 17.9112; % in cm
colwidth = 8.6472; % in cm
% HRTF file name (without the .sofa)
hrirname = 'FABIAN_HRIR_measured_HATO_0';
%% Load HRTF
SOFA_obj = amt_load('engel2022',[hrirname,'.sofa']);
[h,fs,az,el] = sofa2hrtf(SOFA_obj); % get HRTF data in a convenient format
% Zero-pad the HRIRs to increase frequency resolution
taps = 2048;
h = [h;zeros(taps-size(h,1),size(h,2),size(h,3))];
H = ffth(h); % to frequency domain
nfreqs = size(H,1); % number of frequency bins
f = linspace(0,fs/2,nfreqs).'; % vector with frequencies
%% Defining test conditions
ncond = 9;
test_conditions = cell(ncond,1);
% 1. Trunc (no preprocessing, no tapering, no EQ)
test_conditions{1}.name = 'Trunc';
test_conditions{1}.preproc = 'Trunc';
test_conditions{1}.tap = 0;
test_conditions{1}.dualBand = false;
test_conditions{1}.EQ = 0;
% 2. EQ (no preprocessing, no tapering, HRF EQ)
test_conditions{2}.name = 'EQ';
test_conditions{2}.preproc = 'Trunc';
test_conditions{2}.tap = 0;
test_conditions{2}.dualBand = false;
test_conditions{2}.EQ = 2;
% 3. Tap (no preprocessing, dual band tapering with Hann weights, HRF EQ)
test_conditions{3}.name = 'Tap';
test_conditions{3}.preproc = 'Trunc';
test_conditions{3}.tap = 2;
test_conditions{3}.dualBand = true;
test_conditions{3}.EQ = 2;
% 4. TA (time-aligned HRTF, no tapering, no EQ)
test_conditions{4}.name = 'TA';
test_conditions{4}.preproc = 'TA';
test_conditions{4}.tap = 0;
test_conditions{4}.dualBand = false;
test_conditions{4}.EQ = 0;
% 5. MagLS (MagLS, no tapering, no EQ)
test_conditions{5}.name = 'MagLS';
test_conditions{5}.preproc = 'MagLS';
test_conditions{5}.tap = 0;
test_conditions{5}.dualBand = false;
test_conditions{5}.EQ = 0;
% 6. MagLS+CC (MagLS, no tapering, CovConst)
test_conditions{6}.name = 'MagLSCC';
test_conditions{6}.preproc = 'MagLS';
test_conditions{6}.tap = 0;
test_conditions{6}.dualBand = false;
test_conditions{6}.EQ = 1;
% 7. SpSub (SpSub, no tapering, no EQ)
test_conditions{7}.name = 'SpSub';
test_conditions{7}.preproc = 'SpSub';
test_conditions{7}.tap = 0;
test_conditions{7}.dualBand = false;
test_conditions{7}.EQ = 0;
% 8. SpSubMod (SpSub, dual band tapering with Hann weights, HRF EQ)
test_conditions{8}.name = 'SpSubMod';
test_conditions{8}.preproc = 'SpSubMod';
test_conditions{8}.tap = 2;
test_conditions{8}.dualBand = true;
test_conditions{8}.EQ = 2;
% 9. BiMagLS
test_conditions{9}.name = 'BiMagLS';
test_conditions{9}.preproc = 'BiMagLS';
test_conditions{9}.tap = 0;
test_conditions{9}.dualBand = false;
test_conditions{9}.EQ = 0;
%% Define a few direction subsets
% 1. Nearest neighbours to 110-point Lebedev grid
gridTestReq = sofia_lebedev(110,0);
[~,indLeb] = getGridSubset([az,el],gridTestReq,0);
indLeb = find(indLeb); % get indices
azLeb = az(indLeb);
elLeb = el(indLeb);
wLeb = gridTestReq(:,3);
% 2. Median plane (lat = 0 +/- 1 deg)
[lat,~] = sph2hor(az*180/pi,90-el*180/pi); % lat/pol coordinates
indMP = find(abs(lat)<1);
azMP = az(indMP);
elMP = el(indMP);
% 3. Horizontal plane (el = 90 +/- 1 deg)
indHP = find(abs(el-pi/2)<(pi/180));
azHP = az(indHP);
% elHP = el(indHP);
indFront = find(abs(el-pi/2)<(pi/180) & abs(az)<(pi/180)); % front
%% Get high-order HRTF
amt_disp(' ');
amt_disp('Getting reference HRTF...');
hnm_filename = 'hnm_ref';
[Hnm_mag_ref, hnm_ref, Hnm_TA_ref] = amt_cache('get',hnm_filename,flags.cachemode);
if isempty(hnm_ref)
% Standard
hnm_ref = toSH(h,Nref,'az',az,'el',el,'fs',fs,'mode','Trunc');
% Time-aligned
hnm_TA_ref = toSH(h,Nref,'az',az,'el',el,'fs',fs,'mode','TA');
Hnm_TA_ref = ffth(hnm_TA_ref);
% Mag-only
Y = AKsh(Nref,[],az*180/pi,el*180/pi,'real').';
Hnm_mag_ref = mult3(abs(H),pinv(Y));
amt_cache('set',hnm_filename,Hnm_mag_ref,hnm_ref,Hnm_TA_ref,fs);
end
Hnm_ref = ffth(hnm_ref); % to frequency domain
amt_disp('Got reference HRTF!');
%% Get results for high-order HRTF
amt_disp(' ');
amt_disp('Getting results for reference HRTF...');
results_filename = 'res_ref';
results = amt_cache('get', results_filename, flags.cachemode);
if isempty(results)
% --- Numerical analysis --- %
% Magnitude and phase delay 110-point Lebedev grid
results.mag = 20*log10(abs(H(:,indLeb,:)));
results.pd = div2( -unwrap(angle(H(:,indLeb,1)))+unwrap(angle(H(:,indLeb,2))) , 2*pi*f )*1e6;
% Loudness across directions
[L,wERB,calibrationGain] = perceptualSpectrum(H,fs);
Lavgd = sum( mult2(L,wERB) , 1 );
results.L = Lavgd;
results.L_perDirection = L; % save to calculate PSD
results.Lavg = sum( mult2(Lavgd(:,indLeb,:),wLeb.') , 2 ); % weighted avg on the Lebedev grid
results.calibrationGain = calibrationGain; % save to calibrate loudness
% Interaural differences for horizontal plane
results.itd = itdestimator(permute(h(:,indHP,:),[2,3,1]),'fs',fs,...
'MaxIACCe','lp','upper_cutfreq', 3000,...
'butterpoly', 10)*1e6;
results.ild = getILD(h(:,indHP,:),fs);
% --- Reijniers 2014 --- %
amt_disp('Running reijniers2014...');
% Make DTF and SOFA object for Lebedev grid directions only
dtf = getDTF(h(:,indLeb,:),fs);
SOFA_obj = hrtf2sofa(dtf,fs,azLeb,elLeb);
% Preprocessing source information (demo_reijniers2014)
[template_loc, target] = reijniers2014_featureextraction(SOFA_obj);
% Run virtual experiments (demo_reijniers2014)
num_exp = 100;
[doa, ~] = reijniers2014(template_loc, target, 'num_exp', num_exp);
% Calculate performance measures (demo_reijniers2014)
results.lat_acc = reijniers2014_metrics(doa, 'accL'); % mean lateral error
results.lat_prec = reijniers2014_metrics(doa, 'precL'); % lateral std
results.pol_acc = reijniers2014_metrics(doa, 'accP'); % mean polar error
results.pol_prec = reijniers2014_metrics(doa, 'precP'); % polar std
results.template_loc = template_loc; % template DTF
% --- Baumgartner 2021 --- %
amt_disp('Running baumgartner2021...');
% Make DTF for median plane directions only
dtf = getDTF(h(:,indMP,:),fs);
ndirs = numel(elMP);
results.ext = nan(ndirs,1);
template_ext = cell(ndirs,1);
for j=1:ndirs
template_ext{j} = hrtf2sofa(dtf(:,j,:),fs,azMP(j),elMP(j));
% Get externalisation values
results.ext(j) = baumgartner2021(template_ext{j},template_ext{j});
end
results.template_ext = template_ext; % template DTFs
% --- Jelfs 2011 --- %
amt_disp('Running jelfs2011...');
ndirs = numel(azHP);
srm = nan(ndirs,1);
target = squeeze(h(:,indFront,:)); % target fixed at front
for j = 1:ndirs
interferer = squeeze(h(:,indHP(j),:)); % interferer moves around the HP
srm(j) = jelfs2011(target,interferer,fs);
end
results.srm = srm;
% --- Save results --- %
amt_cache('set', results_filename, results);
end
% Some parameters needed for the calculations below:
mag_ref = results.mag; % to calculate magnitude error
pd_ref = results.pd; % to calculate phase delay error
Lref = results.L_perDirection; % to calculate PSD
Lavg_ref = results.Lavg; % to level-match loudness
calibrationGain = results.calibrationGain; % to calibrate perceptual spectrum
template_loc = results.template_loc; % template localisation model
template_ext = results.template_ext; % template externalisation model
clear results
amt_disp('Got results for reference HRTF!');
%% Get results for all low-order HRTFs
amt_disp(' ');
amt_disp('Getting results for all low-order HRTFs...');
nhrtfs = numel(N_vec)*ncond;
hrtfcount = 0;
for N=N_vec % iterate through spatial orders
for i=1:ncond % iterate through conditions
hrtfcount = hrtfcount+1;
name = test_conditions{i}.name;
results_filename = sprintf('res_ord%0.2d_%s',N,name);
results = amt_cache('get',results_filename,flags.cachemode);
if isempty(results) % generate results if not available
amt_disp(' ');
amt_disp(sprintf('Generating HRTF %d/%d...',hrtfcount,nhrtfs));
hnm_filename = sprintf('hnm_ord%0.2d_%s',N,name);
%hnm = amt_cache('get',hnm_filename,flags.cachemode);
%if isempty(hnm) % generate hnm if not available
% amt_disp(['Generating ',num2str(hnm_filename),'...']);
[hnm,fs,varOut] = toSH(h,N,'az',az,'el',el,'fs',fs,...
'mode' ,test_conditions{i}.preproc,...
'tapering',test_conditions{i}.tap,...
'dualBand',test_conditions{i}.dualBand,...
'EQ',test_conditions{i}.EQ,...
'Hnm_ref',Hnm_ref); % reference Hnm only used for EQ
amt_cache('set', hnm_filename, hnm, fs, varOut);
%end
% --- Interpolate HRTF --- %
isaligned = strcmp(name,'TA') || strcmp(name,'BiMagLS');
hInterp = fromSH(hnm,fs,az,el,isaligned,r);
HInterp = ffth(hInterp);
% --- Numerical analysis --- %
% Magnitude and phase delay error 110-point Lebedev grid
mag = 20*log10(abs(HInterp(:,indLeb,:)));
pd = div2( -unwrap(angle(HInterp(:,indLeb,1)))+unwrap(angle(HInterp(:,indLeb,2))) , 2*pi*f )*1e6;
results.err_mag = mean(abs(mag-mag_ref),2); % avg abs difference across directions
results.err_pd = mean(abs(pd-pd_ref),2); % avg abs difference across directions
% Loudness across directions
[L,wERB] = perceptualSpectrum(HInterp,fs,calibrationGain);
Lavgd = sum( mult2(L,wERB) , 1 ); % ERB-weighted avg loudness per direction
Lavg = sum( mult2(Lavgd(:,indLeb,:),wLeb.') , 2 ); % weighted avg on the Lebedev grid
% L = L - Lavg + Lavg_ref; % level-match with reference
L = L - repmat(Lavg - Lavg_ref,size(L,1),size(L,2),1); % level-match with reference
Lavgd = sum( mult2(L,wERB) , 1 ); % recalculate avg loudness per direction
PSD = sum( mult2(abs(L-Lref),wERB) , 1 ); % avg abs difference per direction (PSD)
PSDavg = sum( mult2(PSD(:,indLeb,:),wLeb.') , 2 ); % avg PSD over the Lebedev grid
results.L = Lavgd;
results.PSD = PSD;
results.PSDavg = PSDavg;
% Interaural differences for horizontal plane
results.itd = itdestimator(permute(hInterp(:,indHP,:),[2,3,1]),'fs',fs,...
'MaxIACCe','lp','upper_cutfreq', 3000,...
'butterpoly', 10)*1e6;
results.ild = getILD(hInterp(:,indHP,:),fs);
% --- Reijniers 2014 --- %
amt_disp('Running reijniers2014...');
% Make DTF and SOFA object for Lebedev grid directions only
dtf = getDTF(hInterp(:,indLeb,:),fs);
SOFA_obj = hrtf2sofa(dtf,fs,azLeb,elLeb);
% Preprocessing source information (demo_reijniers2014)
[~, target] = reijniers2014_featureextraction(SOFA_obj);
% Run virtual experiments (demo_reijniers2014)
num_exp = 100;
[doa, ~] = reijniers2014(template_loc, target, 'num_exp', num_exp);
% Calculate performance measures (demo_reijniers2014)
results.lat_acc = reijniers2014_metrics(doa, 'accL'); % mean lateral error
results.lat_prec = reijniers2014_metrics(doa, 'precL'); % lateral std
results.pol_acc = reijniers2014_metrics(doa, 'accP'); % mean polar error
results.pol_prec = reijniers2014_metrics(doa, 'precP'); % polar std
% --- Baumgartner 2021 --- %
amt_disp('Running baumgartner2021...');
% Make DTF for median plane directions only
dtf = getDTF(hInterp(:,indMP,:),fs);
ndirs = numel(elMP);
results.ext = nan(ndirs,1);
for j=1:ndirs
target = hrtf2sofa(dtf(:,j,:),fs,azMP(j),elMP(j));
% Get externalisation values
results.ext(j) = baumgartner2021(target,template_ext{j});
end
% --- Jelfs 2011 --- %
amt_disp('Running jelfs2011...');
ndirs = numel(azHP);
srm = nan(ndirs,1);
target = squeeze(hInterp(:,indFront,:)); % target fixed at front
for j = 1:ndirs
interferer = squeeze(hInterp(:,indHP(j),:)); % interferer moves around the HP
srm(j) = jelfs2011(target,interferer,fs);
end
results.srm = srm;
% --- Save results --- %
amt_cache('set',results_filename,results);
end
end
end
amt_disp('Got results for all low-order HRTFs!');
%% Plot Fig. 1 (SH spectra)
if flags.do_fig1
fig1size = [pagewidth 4.5];
fig1 = figure('units','centimeters','PaperUnits','centimeters',...
'PaperSize',fig1size,'Renderer','painters',...
'position',[2 2 fig1size(1) fig1size(2)],...
'paperposition',[0 0 fig1size(1) fig1size(2)]);
% Tight subplot
gap = [.02 .008]; % gap between subplots in norm units (height width)
marg_h = [.18 .03]; % figure height margins in norm units (lower upper)
marg_w = [.05 .08]; % figure width margins in norm units (left right)
[ha, ~] = tight_subplot(1,3,gap,marg_h,marg_w);
% Fig. 1a: original HRTF
axes(ha(1)), plotSHenergy(Hnm_ref(:,:,1),fs); % left only
xlabel('f (Hz)'), ylabel('Order (n)')
axpos = get(ha(1),'position');
annotation(fig1,'textbox',...
[axpos(1)+axpos(3)-0.2 axpos(2)+axpos(4)-0.09 0.2 0.09],...
'String',{'a) Original'},...
'HorizontalAlignment','right',...
'FontWeight','bold',...
'FontSize',7,...
'FitBoxToText','off',...
'EdgeColor','none');
set(gca,'fontsize',7)
% Fig. 1b: time-aligned HRTF
axes(ha(2)), plotSHenergy(Hnm_TA_ref(:,:,1),fs);
xlabel('f (Hz)'), ylabel('');
set(ha(2),'YTickLabel',{})
axpos = get(ha(2),'position');
annotation(fig1,'textbox',...
[axpos(1)+axpos(3)-0.2 axpos(2)+axpos(4)-0.09 0.2 0.09],...
'String',{'b) Time-aligned'},...
'HorizontalAlignment','right',...
'FontWeight','bold',...
'FontSize',7,...
'FitBoxToText','off',...
'EdgeColor','none');
set(gca,'fontsize',7)
% Fig. 1c: magnitude-only HRTF
axes(ha(3)), plotSHenergy(Hnm_mag_ref(:,:,1),fs);
xlabel('f (Hz)'), ylabel('');
set(ha(3),'YTickLabel',{})
axpos = get(ha(3),'position');
annotation(fig1,'textbox',...
[axpos(1)+axpos(3)-0.2 axpos(2)+axpos(4)-0.09 0.2 0.09],...
'String',{'c) Magnitude only'},...
'HorizontalAlignment','right',...
'FontWeight','bold',...
'FontSize',7,...
'FitBoxToText','off',...
'EdgeColor','none');
set(gca,'fontsize',7)
if isoctave
c = colorbar()
ylabel (c, 'Energy (dB)');
else
c = colorbar; c.Label.String = 'Energy (dB)';
c.Position = [0.9282 0.1784 0.0199 0.7934];
end
end
%% Plot Fig. 2 (mag/phase errors per spatial order)
if flags.do_fig2
% Load data from file
names = {
'ord01_Trunc'
'ord05_Trunc'
'ord10_Trunc'
'ord20_Trunc'
'ord30_Trunc'
'ord44_Trunc'
};
labels = {
'Trunc (N=1)'
'Trunc (N=5)'
'Trunc (N=10)'
'Trunc (N=20)'
'Trunc (N=30)'
'Trunc (N=44)'
};
n = numel(names);
err_mag = zeros(nfreqs,n);
err_pd = zeros(nfreqs,n);
for i=1:n
name = ['res_',names{i}];
results = amt_cache('get', name, flags.cachemode);
err_mag(:,i) = results.err_mag(:,1,1); % left ear only
err_pd(:,i) = results.err_pd;
end
% Tight subplot
fig2size = [colwidth, 8.47];
fig2 = figure('units','centimeters','PaperUnits','centimeters',...
'PaperSize',fig2size,'Renderer','painters',...
'position',[2 2 fig2size(1) fig2size(2)],...
'paperposition',[0 0 fig2size(1) fig2size(2)]);
gap = [.02 .008]; % gap between subplots in norm units (height width)
marg_h = [.09 .02]; % [.09 .1] figure height margins in norm units (lower upper)
marg_w = [.1 .03]; % figure width margins in norm units (left right)
[ha, ~] = tight_subplot(2,1,gap,marg_h,marg_w);
% Top plot: magnitude error
colors = parula(n+1);
lsvec = {'-'};%,'-.'};
lwvec = [0.5,0.5];
mvec = {'^','v','x','s','o','d','p','h','>'};
ms = 3; % marker size
mi = int32(logspace(log10(1),log10(1025),10));
axes(ha(1))
for i=1:n
ls = lsvec{mod(i-1,numel(lsvec))+1};
lw = lwvec(mod(i-1,numel(lwvec))+1);
m = mvec{i};
% For Matlab version >= R2016b:
%semilogx(f,err_mag(:,i),'Color',colors(i,:),'LineWidth',lw,'LineStyle',ls,'Marker',m,'MarkerSize',ms,'MarkerIndices',mi); hold on
% For older versions:
semilogx(f,err_mag(:,i),'Color',colors(i,:),'LineWidth',lw,'LineStyle',ls); hold on
semilogx(f(mi),err_mag(mi,i),'Color',colors(i,:),'LineStyle','none','Marker',m,'MarkerSize',ms,'HandleVisibility','off')
end
semilogx([f(2) 20000], [1 1],'k:')
legend(labels,'location','west')
xlabel(''), xlim([f(2) 20000])
set(ha(1),'XTickLabel',{})
ylabel('Error (dB)'), grid on
axpos = get(ha(1),'position');
annotation(fig2,'textbox',...
[axpos(1) axpos(2)+axpos(4)-0.09 0.5 0.09],...
'String',{'a. Magnitude error'},...
'HorizontalAlignment','left',...
'FontWeight','bold',...
'FontSize',7,...
'FitBoxToText','off',...
'EdgeColor','none');
% Bottom plot: phase error
axes(ha(2))
for i=1:n
ls = lsvec{mod(i-1,numel(lsvec))+1};
lw = lwvec(mod(i-1,numel(lwvec))+1);
m = mvec{i};
% For Matlab version >= R2016b:
%semilogx(f,err_pd(:,i),'Color',colors(i,:),'LineWidth',lw,'LineStyle',ls,'Marker',m,'MarkerSize',ms,'MarkerIndices',mi); hold on
% For older versions:
semilogx(f,err_pd(:,i),'Color',colors(i,:),'LineWidth',lw,'LineStyle',ls); hold on
semilogx(f(mi),err_pd(mi,i),'Color',colors(i,:),'LineStyle','none','Marker',m,'MarkerSize',ms,'HandleVisibility','off')
end
semilogx([f(2) 20000], [20 20],'k:')
xlim([f(2) 20000]), grid on, ylabel('Error (\mus)')
set(ha(2),'XTick',[100,1000,10000,20000],'XTickLabel',{'100','1k','10k','20k'})
xlabel('f (Hz)')
axpos = get(ha(2),'position');
annotation(fig2,'textbox',...
[axpos(1) axpos(2)+axpos(4)-0.09 0.5 0.09],...
'String',{'b. Phase delay error'},...
'HorizontalAlignment','left',...
'FontWeight','bold',...
'FontSize',7,...
'FitBoxToText','off',...
'EdgeColor','none');
set(ha,'fontsize',7)
end
%% Plot Fig. 3 (ITD/ILD per spatial order)
if flags.do_fig3
% Load data from file
names = {
'ord01_Trunc'
'ord05_Trunc'
'ord10_Trunc'
'ord20_Trunc'
'ord30_Trunc'
'ord44_Trunc'
'ref'
};
labels = {
'N=1'
'N=5'
'N=10'
'N=20'
'N=30'
'N=44'
'Reference'
};
n = numel(names);
ndirs = numel(azHP);
itd = zeros(ndirs,n);
ild = zeros(ndirs,n);
for i=1:n
name = ['res_',names{i}];
results = amt_cache('get', name, flags.cachemode);
itd(:,i) = results.itd;
ild(:,i) = results.ild;
end
% Tight subplot
fig3size = [pagewidth 7.3025];
fig3 = figure('units','centimeters','position',[2 2 fig3size(1) fig3size(2)],...
'Renderer','painters','PaperSize',[fig3size(1) fig3size(2)],...
'paperposition',[0 0 fig3size(1) fig3size(2)]);
gap = [.04 .06]; % gap between subplots in norm units (height width)
marg_h = [0 .05]; % [.09 .1] figure height margins in norm units (lower upper)
marg_w = [.03 .23]; % figure width margins in norm units (left right)
[ha, ~] = tight_subplot(1,2,gap,marg_h,marg_w);
colors = parula(n);
lsvec = {'-'};%,'-.'};
lwvec = [0.5,0.5];
mvec = {'^','v','x','s','o','d','p','h','>'};
ms = 3; % marker size
step_big = round(numel(azHP)/4);
step_small = round(step_big/n);
% ITD
axes(ha(1))
for i=1:n % multiply by 0.001 to have it in ms (cleaner plot)
ls = lsvec{mod(i-1,numel(lsvec))+1};
lw = lwvec(mod(i-1,numel(lwvec))+1);
m = mvec{i};
mi = [((i-1)*step_small+1):step_big:numel(azHP)-1]; % marker indices
color = colors(i,:);
if i==n % reference
ls = ':';
lw = 0.5;
m = 'none';
color = [0 0 0];
end
% For Matlab version >= R2016b:
%polarplot(azHP,abs(itd(:,i))*0.001,'Color',color,'LineWidth',lw,'LineStyle',ls,'Marker',m,'MarkerSize',ms,'MarkerIndices',mi), hold on
% For older versions:
polarplot(azHP,abs(itd(:,i))*0.001,'Color',color,'LineWidth',lw,'LineStyle',ls); hold on
polarplot(azHP(mi),abs(itd(mi,i))*0.001,'Color',color,'LineStyle','none','Marker',m,'MarkerSize',ms,'HandleVisibility','off')
end
set(gca,'ThetaZeroLocation','top','ThetaLim',[0 360],'fontsize',7)
set(gca,'ThetaTick',[0:45:360],'ThetaTickLabel',{'0','45','90','135','','-135','-90','-45'},'RAxisLocation',-90)
title('ITD (ms)')
% ILD
axes(ha(2))
for i=1:n
ls = lsvec{mod(i-1,numel(lsvec))+1};
lw = lwvec(mod(i-1,numel(lwvec))+1);
m = mvec{i};
mi = [((i-1)*step_small+1):step_big:numel(azHP)-1]; % marker indices
color = colors(i,:);
if i==n % reference
ls = ':';
lw = 0.5;
m = 'none';
color = [0 0 0];
end
% For Matlab version >= R2016b:
% polarplot(azHP,abs(ild(:,i)),'Color',color,'LineWidth',lw,'LineStyle',ls,'Marker',m,'MarkerSize',ms,'MarkerIndices',mi), hold on
% For older versions:
polarplot(azHP,abs(ild(:,i)),'Color',color,'LineWidth',lw,'LineStyle',ls); hold on
polarplot(azHP(mi),abs(ild(mi,i)),'Color',color,'LineStyle','none','Marker',m,'MarkerSize',ms,'HandleVisibility','off')
end
set(gca,'ThetaZeroLocation','top','ThetaLim',[0 360],'fontsize',7)
set(gca,'ThetaTick',[0:45:360],'ThetaTickLabel',{'0','45','90','135','','-135','-90','-45'},'RAxisLocation',-90)
title('ILD (dB)')
legend(labels,'position',[0.8217 0.3176 0.1331 0.3094]);
% Violins
fig3bsize = [fig3size(1) 3.3];
fig3b = figure('units','centimeters','position',[2 2 fig3bsize(1) fig3bsize(2)],...
'Renderer','painters','PaperSize',[fig3bsize(1) fig3bsize(2)],...
'paperposition',[0 0 fig3bsize(1) fig3bsize(2)]);
gap = [.04 .06]; % gap between subplots in norm units (height width)
marg_h = [0.1 0.03];
marg_w = [0.06 0.01];
[ha, ~] = tight_subplot(1,2,gap,marg_h,marg_w);
itderr = abs(itd-repmat(itd(:,end), 1, size(itd, 2)));
ilderr = abs(ild-repmat(ild(:,end), 1, size(ild, 2)));
axes(ha(1)), violinplot(itderr(:,1:end-1),[],'ShowData',false,'BoxWidth',0.03); grid on
set(gca,'YTickLabelMode','auto')
hold on, plot([0 n],[itdJND itdJND],'k:')
set(gca,'XTickLabel',labels(1:end-1))
ylabel('Abs. ITD error (ms)')
set(gca,'fontsize',7);
axes(ha(2)), violinplot(ilderr(:,1:end-1),[],'ShowData',false,'BoxWidth',0.03); grid on
set(gca,'YTickLabelMode','auto')
hold on, plot([0 n],[ildJND ildJND],'k:')
set(gca,'XTickLabel',labels(1:end-1))
ylabel('Abs. ILD error (dB)')
set(gca,'fontsize',7);
end
%% Plot Fig. 4 (mag/phase errors per method)
if flags.do_fig4
% Load data from file
names = {
'ord03_Trunc'
'ord03_EQ'
'ord03_Tap'
'ord03_TA'
'ord03_MagLS'
'ord03_MagLSCC'
'ord03_SpSub'
'ord03_SpSubMod'
'ord03_BiMagLS'
};
labels = {
'Trunc'
'EQ'
'Tap'
'TA'
'MagLS'
'MagLS+CC'
'SpSub'
'SpSubMod'
'BiMagLS'
};
n = numel(names);
err_mag = zeros(nfreqs,n);
err_pd = zeros(nfreqs,n);
for i=1:n
name = ['res_',names{i}];
results = amt_cache('get', name, flags.cachemode);
err_mag(:,i) = results.err_mag(:,1,1); % left ear only
err_pd(:,i) = results.err_pd;
end
N = 3;
c = 343;
fa = N*c/(2*pi*r); % aliasing frequency
% Tight subplot
fig4size = [colwidth, 8.47];
fig4 = figure('units','centimeters','PaperUnits','centimeters',...
'PaperSize',fig4size,'Renderer','painters',...
'position',[2 2 fig4size(1) fig4size(2)],...
'paperposition',[0 0 fig4size(1) fig4size(2)]);
gap = [.02 .008]; % gap between subplots in norm units (height width)
marg_h = [.09 .02]; % [.09 .1] figure height margins in norm units (lower upper)
marg_w = [.1 .03]; % figure width margins in norm units (left right)
[ha, ~] = tight_subplot(2,1,gap,marg_h,marg_w);
% Top plot: magnitude error
colors = parula(n+1);
lsvec = {'-'};%,'-.'};
lwvec = [0.5,0.5];
mvec = {'^','v','x','s','o','d','p','h','>'};
ms = 3; % marker size
mi = int32(logspace(log10(1),log10(1025),10));
axes(ha(1))
for i=1:n
ls = lsvec{mod(i-1,numel(lsvec))+1};
lw = lwvec(mod(i-1,numel(lwvec))+1);
m = mvec{i};
% For Matlab version >= R2016b:
%semilogx(f,err_mag(:,i),'Color',colors(i,:),'LineWidth',lw,'LineStyle',ls,'Marker',m,'MarkerSize',ms,'MarkerIndices',mi); hold on
% For older versions:
semilogx(f,err_mag(:,i),'Color',colors(i,:),'LineWidth',lw,'LineStyle',ls); hold on
semilogx(f(mi),err_mag(mi,i),'Color',colors(i,:),'LineStyle','none','Marker',m,'MarkerSize',ms,'HandleVisibility','off')
end
xlabel(''), xlim([f(2) 20000])
set(ha(1),'XTickLabel',{})
semilogx([fa fa],get(gca,'ylim'),'k--')
semilogx(get(gca,'xlim'), [1 1],'k:')
legend(labels,'position',[0.1167 0.5817 0.2827 0.3392])
ylabel('Error (dB)'), grid on
axpos = get(ha(1),'position');
annotation(fig4,'textbox',...
[axpos(1) axpos(2)+axpos(4)-0.09 0.5 0.09],...
'String',{'a. Magnitude error'},...
'HorizontalAlignment','left',...
'FontWeight','bold',...
'FontSize',7,...
'FitBoxToText','off',...
'EdgeColor','none');
% Bottom plot: phase error
axes(ha(2))
for i=1:n
ls = lsvec{mod(i-1,numel(lsvec))+1};
lw = lwvec(mod(i-1,numel(lwvec))+1);
m = mvec{i};
% For Matlab version >= R2016b:
%semilogx(f,err_pd(:,i),'Color',colors(i,:),'LineWidth',lw,'LineStyle',ls,'Marker',m,'MarkerSize',ms,'MarkerIndices',mi); hold on
% For older versions:
semilogx(f,err_pd(:,i),'Color',colors(i,:),'LineWidth',lw,'LineStyle',ls); hold on
semilogx(f(mi),err_pd(mi,i),'Color',colors(i,:),'LineStyle','none','Marker',m,'MarkerSize',ms,'HandleVisibility','off')
end
xlim([f(2) 20000]), grid on, ylabel('Error (\mus)')
semilogx([fa fa],get(gca,'ylim'),'k--')
semilogx(get(gca,'xlim'), [20 20],'k:')
set(ha(2),'XTick',[100,1000,10000,20000],'XTickLabel',{'100','1k','10k','20k'})
xlabel('f (Hz)')
axpos = get(ha(2),'position');
annotation(fig4,'textbox',...
[axpos(1) axpos(2)+axpos(4)-0.09 0.5 0.09],...
'String',{'b. Phase delay error'},...
'HorizontalAlignment','left',...
'FontWeight','bold',...
'FontSize',7,...
'FitBoxToText','off',...
'EdgeColor','none');
set(ha,'fontsize',7)
end
%% Plot Fig. 5 (ITD/ILD per method)
if flags.do_fig5
% Load data from file
names = {
'ord03_Trunc'
'ord03_EQ'
'ord03_Tap'
'ord03_TA'
'ord03_MagLS'
'ord03_MagLSCC'
'ord03_SpSub'
'ord03_SpSubMod'
'ord03_BiMagLS'
'ref'
};
labels = {
'Trunc'
'EQ'
'Tap'
'TA'
'MagLS'
'MagLS+CC'
'SpSub'
'SpSubMod'
'BiMagLS'
'Reference'
};
n = numel(names);
ndirs = numel(azHP);
itd = zeros(ndirs,n);
ild = zeros(ndirs,n);
for i=1:n
name = ['res_',names{i}];
results = amt_cache('get', name, flags.cachemode);
itd(:,i) = results.itd;
ild(:,i) = results.ild;
end
% Tight subplot
fig5size = [pagewidth 7.3025];
fig5 = figure('units','centimeters','position',[2 2 fig5size(1) fig5size(2)],...
'Renderer','painters','PaperSize',[fig5size(1) fig5size(2)],...
'paperposition',[0 0 fig5size(1) fig5size(2)]);
gap = [.04 .06]; % gap between subplots in norm units (height width)
marg_h = [0 .05]; % [.09 .1] figure height margins in norm units (lower upper)
marg_w = [.03 .23]; % figure width margins in norm units (left right)
[ha, ~] = tight_subplot(1,2,gap,marg_h,marg_w);
colors = parula(n+1);
lsvec = {'-'};%,'-.'};
lwvec = [0.5,0.5];
mvec = {'^','v','x','s','o','d','p','h','>','+'};
ms = 3; % marker size
step_big = round(numel(azHP)/4);
step_small = round(step_big/n);
% ITD
axes(ha(1))
for i=1:n % multiply by 0.001 to have it in ms (cleaner plot)
ls = lsvec{mod(i-1,numel(lsvec))+1};
lw = lwvec(mod(i-1,numel(lwvec))+1);
m = mvec{i};
mi = [((i-1)*step_small+1):step_big:numel(azHP)-1]; % marker indices
color = colors(i,:);
if i==n % reference
ls = ':';
lw = 0.5;
m = 'none';
color = [0 0 0];
end
% For Matlab version >= R2016b:
%polarplot(azHP,abs(itd(:,i))*0.001,'Color',color,'LineWidth',lw,'LineStyle',ls,'Marker',m,'MarkerSize',ms,'MarkerIndices',mi), hold on
% For older versions:
polarplot(azHP,abs(itd(:,i))*0.001,'Color',color,'LineWidth',lw,'LineStyle',ls); hold on
polarplot(azHP(mi),abs(itd(mi,i))*0.001,'Color',color,'LineStyle','none','Marker',m,'MarkerSize',ms,'HandleVisibility','off')
end
set(gca,'ThetaZeroLocation','top','ThetaLim',[0 360],'fontsize',7)
set(gca,'ThetaTick',[0:45:360],'ThetaTickLabel',{'0','45','90','135','','-135','-90','-45'},'RAxisLocation',-90)
title('ITD (ms)')
% ILD
axes(ha(2))
for i=1:n
ls = lsvec{mod(i-1,numel(lsvec))+1};
lw = lwvec(mod(i-1,numel(lwvec))+1);
m = mvec{i};
mi = [((i-1)*step_small+1):step_big:numel(azHP)-1]; % marker indices
color = colors(i,:);
if i==n % reference
ls = ':';
lw = 0.5;
m = 'none';
color = [0 0 0];
end
% For Matlab version >= R2016b:
%polarplot(azHP,abs(ild(:,i)),'Color',color,'LineWidth',lw,'LineStyle',ls,'Marker',m,'MarkerSize',ms,'MarkerIndices',mi), hold on
% For older versions:
polarplot(azHP,abs(ild(:,i)),'Color',color,'LineWidth',lw,'LineStyle',ls); hold on
polarplot(azHP(mi),abs(ild(mi,i)),'Color',color,'LineStyle','none','Marker',m,'MarkerSize',ms,'HandleVisibility','off')
end
set(gca,'ThetaZeroLocation','top','ThetaLim',[0 360],'fontsize',7)
set(gca,'ThetaTick',[0:45:360],'ThetaTickLabel',{'0','45','90','135','','-135','-90','-45'},'RAxisLocation',-90)
title('ILD (dB)')
legend(labels,'position',[0.8211 0.2543 0.1402 0.4362]);
% Violins
fig5bsize = [fig5size(1) fig5size(2)*0.55];
fig5b = figure('units','centimeters','position',[2 2 fig5bsize(1) fig5bsize(2)],...
'Renderer','painters','PaperSize',[fig5bsize(1) fig5bsize(2)],...
'paperposition',[0 0 fig5bsize(1) fig5bsize(2)]);
gap = [.04 .06]; % gap between subplots in norm units (height width)
marg_h = [0.3 0.03];
marg_w = [0.06 0.01];
[ha, ~] = tight_subplot(1,2,gap,marg_h,marg_w);
itderr = abs(itd-repmat(itd(:,end), 1, size(itd, 2)));
ilderr = abs(ild-repmat(ild(:,end), 1, size(ild, 2)));
axes(ha(1)), violinplot(itderr(:,1:end-1),[],'ShowData',false,'BoxWidth',0.03); grid on
set(gca,'YTickLabelMode','auto')
hold on, plot([0 n],[itdJND itdJND],'k:')
set(ha(1),'XTickLabel',labels(1:end-1))
ylabel('Abs. ITD error (ms)')
set(gca,'fontsize',7,'xticklabelrotation',45)
axes(ha(2)), violinplot(ilderr(:,1:end-1),[],'ShowData',false,'BoxWidth',0.03); grid on
set(gca,'YTickLabelMode','auto')
hold on, plot([0 n],[ildJND ildJND],'k:')
set(ha(2),'XTickLabel',labels(1:end-1))
ylabel('Abs. ILD error (dB)')
set(gca,'fontsize',7,'xticklabelrotation',45)
end
%% Plot Fig. 6 (loudness per direction)
if flags.do_fig6
% Load data from file
names = {
'ref'
'ord03_Trunc'
'ord03_EQ'
'ord03_Tap'
'ord03_TA'
'ord03_MagLS'
'ord03_MagLSCC'
'ord03_SpSub'
'ord03_SpSubMod'
'ord03_BiMagLS'
};
labels = {
'Reference'
'Trunc'
'EQ'
'Tap'
'TA'
'MagLS'
'MagLS+CC'
'SpSub'
'SpSubMod'
'BiMagLS'
};
n = numel(names);
ndirs = numel(az);
L = cell(n,1);
PSD = zeros(ndirs,n);
PSDavg = zeros(n,1);
for i=1:n
name = ['res_',names{i}];
results = amt_cache('get', name, flags.cachemode);
L{i} = results.L;
if isfield(results,'PSD')
PSD(:,i) = results.PSD(:,:,1).'; % left ear only
PSDavg(i) = results.PSDavg(:,:,1); % left ear only
end
end
% Tight subplot
fig6size = [pagewidth 6.9003];
fig6 = figure('units','centimeters','position',[2 2 fig6size(1) fig6size(2)],...
'Renderer','painters','PaperSize',[fig6size(1) fig6size(2)],...
'paperposition',[0 0 fig6size(1) fig6size(2)]);
gap = [.07 .008]; % gap between subplots in norm units (height width)
marg_h = [.13 .06]; % [.09 .1] figure height margins in norm units (lower upper)
marg_w = [.06 .08]; % figure width margins in norm units (left right)
[ha, ~] = tight_subplot(2,5,gap,marg_h,marg_w);
clims = [1 5];
colormap parula
for i=1:n
axes(ha(i))
plotSph(az,el,L{i}(:,:,1)) % left ear only
if i==1 || i==6
set(ha(i),'YTick',-60:30:60)
ylabel('Elevation (deg)')
else
set(ha(i),'YTickLabel',{})
end
if i>=6
xlabel('Azimuth (deg)')
else
set(ha(i),'XTickLabel',{})
end
title(labels{i})
set(gca,'fontsize',7,'XDir','reverse')
grid(gca,'off')
caxis(clims)
if i==n
c = colorbar; c.Label.String = 'Loudness (sones)';
c.Position = [0.9373 0.3497 0.0111 0.3804];
end
end
for i=(n+1):10
axis(ha(i),'off')
end
% Violins
fig6bsize = [fig6size(1) fig6size(2)/2];
fig6b = figure('units','centimeters','position',[2 2 fig6bsize(1) fig6bsize(2)],...
'Renderer','painters','PaperSize',[fig6bsize(1) fig6bsize(2)],...
'paperposition',[0 0 fig6bsize(1) fig6bsize(2)]);
violinplot(PSD(:,2:end),[],'ShowData',false,'BoxWidth',0.03); grid on
set(gca,'XTickLabel',labels(2:end))
ylabel('PSD (sones)')
set(gca,'fontsize',7)
end
%% Plot Fig. 7 (models' outputs per spatial order)
if flags.do_fig7
% Load data from file
names = {
'Trunc'
'EQ'
'Tap'
'TA'
'MagLS'
'MagLSCC'
'SpSub'
'SpSubMod'
'BiMagLS'
'ref'
};
labels = {
'Trunc'
'EQ'
'Tap'
'TA'
'MagLS'
'MagLS+CC'
'SpSub'
'SpSubMod'
'BiMagLS'
'Reference'
};
n = numel(names);
m = numel(N_vec);
PSD = nan(n,m);
lat_prec = nan(n,m);
pol_prec = nan(n,m);
ext = nan(n,m);
srm = nan(n,m);
for i=1:n
name = names{i};
if strcmp(name,'ref')
results = amt_cache('get', ['res_',name], flags.cachemode);
PSD(i,:) = 0;
lat_prec(i,:) = results.lat_prec;
pol_prec(i,:) = results.pol_prec;
ext(i,:) = mean(results.ext);
srm(i,:) = mean(results.srm);
else
for j=1:m
N=N_vec(j);
results = amt_cache('get', sprintf('res_ord%0.2d_%s',N,name), flags.cachemode);
PSD(i,j) = results.PSDavg(:,:,1); % left ear only
lat_prec(i,j) = results.lat_prec;
pol_prec(i,j) = results.pol_prec;
ext(i,j) = mean(results.ext);
srm(i,j) = mean(results.srm);
end
end
end
% Tight subplot
fig7size = [pagewidth 9.7790];
fig7 = figure('units','centimeters','position',[2 2 fig7size(1) fig7size(2)],...
'Renderer','painters','PaperSize',[fig7size(1) fig7size(2)],...
'paperposition',[0 0 fig7size(1) fig7size(2)]);
gap = [.06 .05]; % gap between subplots in norm units (height width)
marg_h = [.1 .03]; % [.09 .1] figure height margins in norm units (lower upper)
marg_w = [.07 .02]; % figure width margins in norm units (left right)
[ha, ~] = tight_subplot(2,3,gap,marg_h,marg_w);
colors = parula(n);
lsvec = {'-'};%,'-.'};
mvec = {'^','v','x','s','o','d','p','h','>'};
lwvec = [0.5,0.5];
ms = 3; % marker size
mi = [1:5:44]; % marker indices
colors(end,:) = [0 0 0]; % last one is reference
for i=1:n
if i<n
ls = lsvec{mod(i-1,numel(lsvec))+1};
lw = lwvec(mod(i-1,numel(lwvec))+1);
m = mvec{i};
else
ls = ':'; % reference
lw = 0.5;
m = 'None';
end
% For Matlab version >= R2016b:
%plot(ha(1),PSD(i,:),'Color',colors(i,:),'LineWidth',lw,'LineStyle',ls,'Marker',m,'MarkerSize',ms,'MarkerIndices',mi),hold(ha(1),'on')
%plot(ha(2),lat_prec(i,:),'Color',colors(i,:),'LineWidth',lw,'LineStyle',ls,'Marker',m,'MarkerSize',ms,'MarkerIndices',mi),hold(ha(2),'on')
%plot(ha(3),pol_prec(i,:),'Color',colors(i,:),'LineWidth',lw,'LineStyle',ls,'Marker',m,'MarkerSize',ms,'MarkerIndices',mi),hold(ha(3),'on')
%plot(ha(4),ext(i,:),'Color',colors(i,:),'LineWidth',lw,'LineStyle',ls,'Marker',m,'MarkerSize',ms,'MarkerIndices',mi),hold(ha(4),'on')
%plot(ha(5),srm(i,:),'Color',colors(i,:),'LineWidth',lw,'LineStyle',ls,'Marker',m,'MarkerSize',ms,'MarkerIndices',mi),hold(ha(5),'on')
% For older versions:
plot(ha(1),N_vec,PSD(i,:),'Color',colors(i,:),'LineWidth',lw,'LineStyle',ls),hold(ha(1),'on')
plot(ha(2),N_vec,lat_prec(i,:),'Color',colors(i,:),'LineWidth',lw,'LineStyle',ls),hold(ha(2),'on')
plot(ha(3),N_vec,pol_prec(i,:),'Color',colors(i,:),'LineWidth',lw,'LineStyle',ls),hold(ha(3),'on')
plot(ha(4),N_vec,ext(i,:),'Color',colors(i,:),'LineWidth',lw,'LineStyle',ls),hold(ha(4),'on')
plot(ha(5),N_vec,srm(i,:),'Color',colors(i,:),'LineWidth',lw,'LineStyle',ls),hold(ha(5),'on')
plot(ha(1),N_vec(mi),PSD(i,mi),'Color',colors(i,:),'Marker',m,'MarkerSize',ms,'LineStyle','none','HandleVisibility','off')
plot(ha(2),N_vec(mi),lat_prec(i,mi),'Color',colors(i,:),'Marker',m,'MarkerSize',ms,'LineStyle','none','HandleVisibility','off')
plot(ha(3),N_vec(mi),pol_prec(i,mi),'Color',colors(i,:),'Marker',m,'MarkerSize',ms,'LineStyle','none','HandleVisibility','off')
plot(ha(4),N_vec(mi),ext(i,mi),'Color',colors(i,:),'Marker',m,'MarkerSize',ms,'LineStyle','none','HandleVisibility','off')
plot(ha(5),N_vec(mi),srm(i,mi),'Color',colors(i,:),'Marker',m,'MarkerSize',ms,'LineStyle','none','HandleVisibility','off')
end
legend(ha(1),labels,'position',[0.7678 0.1414 0.1367 0.2944]);
ylabel(ha(1),'PSD (sones)')
ylabel(ha(2),'Lateral precision (deg)')
ylabel(ha(3),'Polar precision (deg)')
ylabel(ha(4),'Externalisation')
ylabel(ha(5),'SRM (dB)')
for i=1:5
grid(ha(i),'on')
xlim(ha(i),[1 44])
set(ha(i),'XTick',[1,5:5:44])
if i>3
xlabel(ha(i),'Spatial order (N)')
end
set(ha(i),'fontsize',7)
end
set(ha(6),'visible','off') % hide last axis
set(ha(1),'yscale','log') % set some plots' Y axis to log scale
set(ha(2),'yscale','log','YTick',[1:5,7,10:10:50,70,100])
set(ha(3),'yscale','log','YTick',[1:5,7,10:10:50,70,100])
end
end
Build with Bootstrap