function varargout=exp_chen2011(varargin)
%EXP_CHEN2011 experiments from Chen et al. 2011
%
% `exp_chen2011(fig)` reproduce Fig. no. *fig* from the Chen et
% al. 2011 paper.
%
% The following flags can be specified;
%
% 'fig7' Reproduce Fig. 7, yields the excitation patterns
%
% 'fig9' Reproduce Fig. 9, contrasting loudness (sones)
% with input level (dBSPL)
%
% 'fig10' Reproduce Fig. 10, comparing the equal loudness contours
% from chen2011_loudness with the ISO 2006 standard
%
% 'fig11' Reproduce Fig. 11, displaying loudness as a function of bandwidth
%
%
% Examples:
% ---------
%
% To display Fig. 7 use :::
%
% exp_chen2011('fig7');
%
% To display Fig. 9 use :::
%
% exp_chen2011('fig9');
%
% To display Fig. 10 use :::
%
% exp_chen2011('fig10');
%
% To display Fig. 11 use :::
%
% exp_chen2011('fig11');
%
% References: chen2011
definput.flags.type = {'missingflag',...
'fig7','fig9','fig10','fig11'};
definput.flags.plot = {'plot','no_plot'};
% 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;
if flags.do_fig7 || flags.do_fig9
OuterEarOpt = 'FreeField';
CF1k = 1000;
% LdB1k = [2 10:10:90]';
LdB1k = [2 5 10:5:90]';
%% ANSI S3.4-2007
phon_ref = [0 1 2 3 4 5 7.5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 105 110 115 120];
sone_ref = [0.0011 0.0018 0.0028 0.0044 0.0065 0.0077 0.017 0.029 0.07 0.142 0.255 0.422 0.662 0.997 1.461 ...
2.098 2.970 4.166 5.813 8.102 11.326 15.98 22.929 33.216 48.242 70.362 103.274 152.777 227.855 341.982];
Ldn1k_Binaural_ANSI = interp1(phon_ref,sone_ref,LdB1k,'linear','extrap');
%% model calculation
for j = 1:length(LdB1k)
[Ldn1k_Monaural(j),E1k(:,j),Cam,CF] = chen2011(CF1k,LdB1k(j), OuterEarOpt);
end
Ldn1k_Binaural_Model = 2* Ldn1k_Monaural;
if flags.do_plot && flags.do_fig9
figure;
semilogy(LdB1k,Ldn1k_Binaural_ANSI,'k--',LdB1k,Ldn1k_Binaural_Model,'k');
legend('ANSI S3.4-2007','This model');
xlabel('Input level of 1-kHz tone, dB SPL');ylabel('Loundess, sones');
axis([0 100 0.001 1000]);
set(findobj('FontSize',10),'FontSize',12);
set(get(gca,'XLabel'),'FontSize',12);
set(get(gca,'YLabel'),'FontSize',12);
set(findobj(get(gca,'Children'),'LineWidth',0.5),'LineWidth',2);
set(gca,'LineWidth',1');
elseif flags.do_plot && flags.do_fig7
figure;
semilogx(CF,10*log10(E1k));
xlabel('CF, Hz');ylabel('Excitation, dB');
axis([20 20000 0 100]);
set(findobj('FontSize',10),'FontSize',12);
set(get(gca,'XLabel'),'FontSize',12);
set(get(gca,'YLabel'),'FontSize',12);
set(findobj(get(gca,'Children'),'LineWidth',0.5),'LineWidth',2);
set(gca,'LineWidth',1');
end
elseif flags.do_fig10
OuterEarOpt = 'FreeField';
%% 1-kHz as reference
CF1k = 1000;
LdB1k = [2 10:10:90]';
for j = 1:length(LdB1k)
Ldn1k(j) = chen2011(CF1k,LdB1k(j), OuterEarOpt);
end
%% calculating equal loudness contours
%this is the data from iso226-2003
data = amt_load('chen2011', 'data_chen2011.mat');
freq = data.freq;
spl = data.spl;
f = freq(:,1);
LdB = -10:2:140;
for i = 1:length(f)
for j = 1:length(LdB)
[Ldn(j),Exitation,Cam,CF] = chen2011(f(i),LdB(j), OuterEarOpt);
end
LdB_sameLdn(i,:) = interp1(Ldn,LdB,Ldn1k,'linear','extrap');
end
if flags.do_plot
figure;
semilogx(freq(:,1),spl(:,1)-1.6,'k--',freq(:,1),LdB_sameLdn(:,1),'k',...
freq(:,2:end),spl(:,2:end),'k--',freq(:,2:end),LdB_sameLdn(:,2:end),'k');
legend('ISO 226-2003','This paper');
xlabel('Frequency, Hz');ylabel('LdB, dB SPL');
axis([20 20000 -10 130]);
set(findobj('FontSize',10),'FontSize',12);
set(get(gca,'XLabel'),'FontSize',12);
set(get(gca,'YLabel'),'FontSize',12);
set(findobj(get(gca,'Children'),'LineWidth',0.5),'LineWidth',2);
set(gca,'LineWidth',1');
text(600,44,'40 phons');
end
elseif flags.do_fig11
%% oringinal data
%%%%%% Fig 14 in Moore et al,1997
NoiseF0 = 1420;
NoiseBw0 = 210;
NoiseLdB01 = 30;
NoiseBw11 = [30 70 150 250 500 1000 3600];
NoiseLdB11 = [30.9 31.7 30.1 30.9 31.7 34.0 34.0];
NoiseLdB02 = 50;
NoiseBw12 = [35 75 150 250 500 1125 3400];
NoiseLdB12 = [51.1 50.3 50.3 50.3 54.2 57.3 61.9];
NoiseLdB03 = 50;
NoiseBw13 = [35 75 150 250 500 1125 3400];
NoiseLdB13 = [79.0 79.8 79.8 79.3 81.4 83.1 86.2];
%% model calculation
OuterEarOpt = 'PDR10';
F0 = 1420; % Hz
%%%%%%%%%%%%%%%% Step1: loudness of matching noise
FixNoiseBW = 210; % Hz
MatchNoiseLdB = 20:5:100;
for k = 1:length(MatchNoiseLdB)
FixNoiseSpectrumLevel(k) = MatchNoiseLdB(k) - 10*log10(FixNoiseBW); % dB SPL/Hz
fl2 = (-FixNoiseBW+sqrt(FixNoiseBW.^2+4*F0^2))/2;
fu2 = fl2 + FixNoiseBW-1;
comp_f = fl2:fu2;
comp_dB = FixNoiseSpectrumLevel(k) * ones(1,length(comp_f));
Ldn2(k) = chen2011(comp_f,comp_dB, OuterEarOpt);
end
%%%%%%%%%%%%%%%% Step2: loudness of variable bandwidth noise
FlexNoiseBW = [25:50:125,150:50:300, 400:100:1000,1250,1500,2000,3000,4000]; % Hz
NoiseLevel = [30 50 80]; % dB SPL
for i = 1:length(NoiseLevel)
for j = 1:length(FlexNoiseBW)
FlexNoiseSpectrumLevel(i,j) = NoiseLevel(i) - 10*log10(FlexNoiseBW(j)); % dB SPL/Hz
fl1 = (-FlexNoiseBW(j)+sqrt(FlexNoiseBW(j).^2+4*F0^2))/2;
fu1 = fl1 + FlexNoiseBW(j)-1;
comp_f = fl1:fu1;
comp_dB = FlexNoiseSpectrumLevel(i,j) * ones(1,length(comp_f));
Ldn1(j) = chen2011(comp_f,comp_dB, OuterEarOpt);
end
LdB_samLdn(i,:) = interp1(Ldn2,MatchNoiseLdB,Ldn1,'linear','extrap');
end
if flags.do_plot
figure;
semilogx(FlexNoiseBW,LdB_samLdn','k',NoiseBw11,NoiseLdB11,'ko',...
NoiseBw12,NoiseLdB12,'ko',NoiseBw13,NoiseLdB13,'ko');
axis([20 5000 20 100]);
xlabel('Variable bandwidth, Hz');ylabel('Level of matching noise, dB SPL');
set(findobj('FontSize',10),'FontSize',12);
set(get(gca,'XLabel'),'FontSize',12,'Vertical','top');
set(get(gca,'YLabel'),'FontSize',12,'Vertical','middle');
set(findobj(get(gca,'Children'),'LineWidth',0.5),'LineWidth',2);
set(gca,'LineWidth',1');
end
end