THE AUDITORY MODELING TOOLBOX
Go to function
EXP_LI2020 - experimental results for externalization model from Li et al. (2020)
Program code:
function exp_li2020(varargin)
%EXP_LI2020 experimental results for externalization model from Li et al. (2020)
%
% EXP_LI2020(flag) demonstrates how to apply the model to estimate the degree of externalization
% by comparing template and target HRTFs
%
% Examples:
% ---------
% To display measured and predicted results of experiment A use :
% exp_li2020('fig2');
%
% To display measured and predicted results of experiment B use :
% exp_li2020('fig3');
%
% To display measured and predicted results of experiment C use :
% exp_li2020('fig5');
%
% To display measured and predicted results of experiment D use :
% exp_li2020('fig6');
%
% To display measured and predicted results of experiment E use :
% exp_li2020('fig7');
%
% Requirements:
% -------------
%
% 1) Data in hrtf/li2020 and auxdata/li2020 (noise_input,...
% ExpResults_LI2020, modified_BRIR_HRTF_dataset)
%
% References:
% S. Li, R. Baumgartner, and J. Peissig. Modeling perceived
% externalization of a static, lateral sound image. Acta Acustica, 4(5),
% 2021.
%
%
% Url: http://amtoolbox.org/amt-1.1.0/doc/experiments/exp_li2020.php
% Copyright (C) 2009-2021 Piotr Majdak, Clara Hollomey, and the AMT team.
% This file is part of Auditory Modeling Toolbox (AMT) version 1.1.0
%
% 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: Song Li, Institute of Communications Technology, Leibniz
% University of Hannover, Germany
% To make the scripts clear, the general weighting factors are used for
% calculating the fitted and predicted reults. The fitted results are
% slightly different than the results in the paper, since the fitted results
% shown in the paper are calcualted based on the individual weighting
% factors.
%% check input
definput.keyvals.Obj = [];
definput.flags.experiment = {'fig2','fig3','fig5','fig6','fig7'};
[flags,kv]=ltfatarghelper({'Obj'},definput,varargin);
%% Get listener's data
%stimulus = load('AMT Implementation\data\noise_input.mat');
stimulus = amt_load('li2020', 'noise_input.mat');
data = data_li2020('hrtf'); % load HRTFs of listener pool
Nsubj = 5; % 5 listeners
fs = 44100;
%% Run model to get externalization scores for every conditions/experiment
if flags.do_fig2
%% Experiment 1
ILD_expansion = 5;
E_predict_exp1=zeros(5,5,3); % 5 subjects x 5 ILD expansion levels x 3 conditions
for ids = 1:Nsubj
% condition 1: broadband BB
for k =1:ILD_expansion
template = squeeze(data.hrir_bb_modified_Exp1(ids,:,1,:));
target = squeeze(data.hrir_bb_modified_Exp1(ids,:,k,:));
[E_predict_exp1(ids,k,1)] = li2020(target,template,[],[],'stim',stimulus.noise_input,'fs',fs,'fsstim',fs, ...
'flow',200, 'fhigh',16000, 'space',1);
end
% condition 2: Low frequency LO
for k =1:ILD_expansion
template = squeeze(data.hrir_low_modified_Exp1(ids,:,1,:));
target = squeeze(data.hrir_low_modified_Exp1(ids,:,k,:));
[E_predict_exp1(ids,k,2)] = li2020(target,template,[],[],'stim',stimulus.noise_input,'fs',fs,'fsstim',fs, ...
'flow',200, 'fhigh',16000, 'space',1);
end
% condition 3: High frequency HI
for k =1:ILD_expansion
template = squeeze(data.hrir_high_modified_Exp1(ids,:,1,:));
target = squeeze(data.hrir_high_modified_Exp1(ids,:,k,:));
[E_predict_exp1(ids,k,3)] = li2020(target,template,[],[],'stim',stimulus.noise_input,'fs',fs,'fsstim',fs, ...
'flow',200, 'fhigh',16000, 'space',1);
end
end
CI95 = [ bsxfun(@times, std(E_predict_exp1(:,:,1))/sqrt(length(E_predict_exp1(:,:,1))),...
tinv([0.025 0.975], length(E_predict_exp1(:,:,1))-1)');
bsxfun(@times, std(E_predict_exp1(:,:,2))/sqrt(length(E_predict_exp1(:,:,2))),...
tinv([0.025 0.975], length(E_predict_exp1(:,:,2))-1)');
bsxfun(@times, std(E_predict_exp1(:,:,3))/sqrt(length(E_predict_exp1(:,:,3))),...
tinv([0.025 0.975], length(E_predict_exp1(:,:,3))-1)')];
CI95 = CI95(2:2:6,:);
data_li2020('exp1','plot');
x_conditions = 1:5;
legend off
errorbar(x_conditions-0.05, mean(E_predict_exp1(:,:,1)), CI95(1,:),'r','LineStyle','none','LineWidth',1.5); hold on;
errorbar(x_conditions, mean(E_predict_exp1(:,:,2)), CI95(2,:),'r','LineStyle','none','LineWidth',1.5);
errorbar(x_conditions+0.05, mean(E_predict_exp1(:,:,3)), CI95(3,:),'r','LineStyle','none','LineWidth',1.5);
plot(x_conditions-0.05, median(E_predict_exp1(:,:,1)),'ro--', 'LineWidth',1.5, 'MarkerSize',10, 'MarkerFaceColor','r');
plot(x_conditions,median(E_predict_exp1(:,:,2)),'rs--', 'LineWidth',1.5, 'MarkerSize',10, 'MarkerFaceColor','r');
plot(x_conditions+0.05,median(E_predict_exp1(:,:,3)),'rd--', 'LineWidth',1.5, 'MarkerSize',10, 'MarkerFaceColor','r');
end
if flags.do_fig3
%% Experiment 2
bandwidth = 5;
E_predict_exp2=zeros(5,5); % 5 subjects x 5 conditions
for ids = 1:Nsubj
for k =1:bandwidth
template = squeeze(data.hrir_modified_Exp2(ids,:,1,:));
target = squeeze(data.hrir_modified_Exp2(ids,:,k,:));
[E_predict_exp2(ids,k)] = li2020(target,template,[],[],'stim',stimulus.noise_input,'fs',fs,'fsstim',fs, ...
'flow',200, 'fhigh',16000, 'space',1);
end
end
CI95 = bsxfun(@times, std(E_predict_exp2)/sqrt(length(E_predict_exp2)),...
tinv([0.025 0.975], length(E_predict_exp2)-1)');
data_li2020('exp2','plot');
x_conditions = 1:5;
legend off
errorbar(x_conditions, mean(E_predict_exp2), CI95(1,:),'r','LineStyle','none','LineWidth',1.5); hold on;
plot(x_conditions,median(E_predict_exp2),'ro--', 'LineWidth',1.5, 'MarkerSize',10, 'MarkerFaceColor','r');
end
if flags.do_fig5
%% Experiment 3
compressed_ILD = 5;
E_predict_exp3=zeros(5,5,2); % 5 subjects x 5 ILD compression levels x 2 conditions
for ids = 1:Nsubj
% condition 1: ipsi
for k =1:compressed_ILD
template = squeeze(data.hrir_ipsi_modified_Exp3(ids,:,1,:));
target = squeeze(data.hrir_ipsi_modified_Exp3(ids,:,k,:));
[E_predict_exp3(ids,k,1)] = li2020(target,template,[],[],'stim',stimulus.noise_input,'fs',fs,'fsstim',fs, ...
'flow',200, 'fhigh',16000, 'space',1);
end
% condition 2: contra
for k =1:compressed_ILD
template = squeeze(data.hrir_contra_modified_Exp3(ids,:,1,:));
target = squeeze(data.hrir_contra_modified_Exp3(ids,:,k,:));
[E_predict_exp3(ids,k,2)] = li2020(target,template,[],[],'stim',stimulus.noise_input,'fs',fs,'fsstim',fs, ...
'flow',200, 'fhigh',16000, 'space',1);
end
end
CI95 = [ bsxfun(@times, std(E_predict_exp3(:,:,1))/sqrt(length(E_predict_exp3(:,:,1))),...
tinv([0.025 0.975], length(E_predict_exp3(:,:,1))-1)');
bsxfun(@times, std(E_predict_exp3(:,:,2))/sqrt(length(E_predict_exp3(:,:,2))),...
tinv([0.025 0.975], length(E_predict_exp3(:,:,2))-1)')];
CI95 = CI95(2:2:4,:);
data_li2020('exp3','plot');
x_conditions = 1:5;
legend off
errorbar(x_conditions-0.05, mean(E_predict_exp3(:,:,1)), CI95(1,:),'r','LineStyle','none','LineWidth',1.5); hold on;
errorbar(x_conditions+0.05, mean(E_predict_exp3(:,:,2)), CI95(2,:),'r','LineStyle','none','LineWidth',1.5);
plot(x_conditions-0.05,median(E_predict_exp3(:,:,1)),'ro--', 'LineWidth',1.5, 'MarkerSize',10, 'MarkerFaceColor','r');
plot(x_conditions+0.05,median(E_predict_exp3(:,:,2)),'rs--', 'LineWidth',1.5, 'MarkerSize',10, 'MarkerFaceColor','r');
end
if flags.do_fig6
%% Experiment 4
bandwidth = 5;
E_predict_exp4=zeros(5,5,3); % 5 subjects x 5 smoothing levels x 3 conditions
for ids = 1:Nsubj
% condition 1: bilateral
for k =1:bandwidth
template = squeeze(data.hrir_both_modified_Exp4(ids,:,1,:));
target = squeeze(data.hrir_both_modified_Exp4(ids,:,k,:));
[E_predict_exp4(ids,k,1)] = li2020(target,template,[],[],'stim',stimulus.noise_input,'fs',fs,'fsstim',fs, ...
'flow',200, 'fhigh',16000, 'space',1);
end
% condition 2: ipsilateral
for k =1:bandwidth
template = squeeze(data.hrir_ipsi_modified_Exp4(ids,:,1,:));
target = squeeze(data.hrir_ipsi_modified_Exp4(ids,:,k,:));
[E_predict_exp4(ids,k,2)] = li2020(target,template,[],[],'stim',stimulus.noise_input,'fs',fs,'fsstim',fs, ...
'flow',200, 'fhigh',16000, 'space',1);
end
% condition 3: contralateral
for k =1:bandwidth
template = squeeze(data.hrir_contra_modified_Exp4(ids,:,1,:));
target = squeeze(data.hrir_contra_modified_Exp4(ids,:,k,:));
[E_predict_exp4(ids,k,3)] = li2020(target,template,[],[],'stim',stimulus.noise_input,'fs',fs,'fsstim',fs, ...
'flow',200, 'fhigh',16000, 'space',1);
end
end
CI95 = [ bsxfun(@times, std(E_predict_exp4(:,:,1))/sqrt(length(E_predict_exp4(:,:,1))),...
tinv([0.025 0.975], length(E_predict_exp4(:,:,1))-1)');
bsxfun(@times, std(E_predict_exp4(:,:,2))/sqrt(length(E_predict_exp4(:,:,2))),...
tinv([0.025 0.975], length(E_predict_exp4(:,:,2))-1)');
bsxfun(@times, std(E_predict_exp4(:,:,3))/sqrt(length(E_predict_exp4(:,:,3))),...
tinv([0.025 0.975], length(E_predict_exp4(:,:,3))-1)')];
CI95 = CI95(2:2:6,:);
data_li2020('exp4','plot');
x_conditions = 1:5;
legend off
errorbar(x_conditions-0.05, mean(E_predict_exp4(:,:,1)), CI95(1,:),'r','LineStyle','none','LineWidth',1.5); hold on
errorbar(x_conditions, mean(E_predict_exp4(:,:,2)), CI95(2,:),'r','LineStyle','none','LineWidth',1.5);
errorbar(x_conditions+0.05, mean(E_predict_exp4(:,:,3)), CI95(3,:),'r','LineStyle','none','LineWidth',1.5);
plot(x_conditions-0.05,median(E_predict_exp4(:,:,1)),'ro--', 'LineWidth',1.5, 'MarkerSize',10, 'MarkerFaceColor','r');
plot(x_conditions,median(E_predict_exp4(:,:,2)),'rs--', 'LineWidth',1.5, 'MarkerSize',10, 'MarkerFaceColor','r');
plot(x_conditions+0.05,median(E_predict_exp4(:,:,3)),'rd--', 'LineWidth',1.5, 'MarkerSize',10, 'MarkerFaceColor','r');
end
if flags.do_fig7
%% Experiment 5
reverbLevel = 5;
% reverbLevel =1: original BRIRs
% reverbLevel =5: Anechoic condition/HRTF
E_predict_exp5=zeros(5,5,5); % 5 subjects x 5 bandwidth factors x 5 reverb levels
for ids = 1:Nsubj
% condition 1: B = 0
for k =1:reverbLevel
template = squeeze(data.brir_both_modified_Exp5(ids,:,1,1,:)); % with given BRIRs
target = squeeze(data.brir_both_modified_Exp5(ids,:,1,k,:));
template_dir = squeeze(data.brir_both_modified_Exp5(ids,:,1,reverbLevel,:)); % with given direct sound part
target_dir = squeeze(data.brir_both_modified_Exp5(ids,:,1,reverbLevel,:));
[E_predict_exp5(ids,1,k)] = li2020(target_dir,template_dir,target,template,'stim',stimulus.noise_input,'fs',fs,'fsstim',fs, ...
'flow',200, 'fhigh',16000, 'space',1);
% [E_predict_exp5(ids,1,k)] = Li2020([],[],target,template,'stim',stimulus.noise_input,'fs',fs,'fsstim',fs, ...
% 'flow',200, 'fhigh',16000, 'space',1); % option
end
% condition 2: B = 1
for k =1:reverbLevel
template = squeeze(data.brir_both_modified_Exp5(ids,:,1,1,:)); % with given BRIRs
target = squeeze(data.brir_both_modified_Exp5(ids,:,2,k,:));
template_dir = squeeze(data.brir_both_modified_Exp5(ids,:,1,reverbLevel,:)); % with given direct sound part
target_dir = squeeze(data.brir_both_modified_Exp5(ids,:,2,reverbLevel,:));
[E_predict_exp5(ids,2,k)] = li2020(target_dir,template_dir,target,template,'stim',stimulus.noise_input,'fs',fs,'fsstim',fs, ...
'flow',200, 'fhigh',16000, 'space',1);
% [E_predict_exp5(ids,2,k)] = Li2020([],[],target,template,'stim',stimulus.noise_input,'fs',fs,'fsstim',fs, ...
% 'flow',200, 'fhigh',16000, 'space',1); % option
end
% condition 3: B = 4
for k =1:reverbLevel
template = squeeze(data.brir_both_modified_Exp5(ids,:,1,1,:)); % with given BRIRs
target = squeeze(data.brir_both_modified_Exp5(ids,:,3,k,:));
template_dir = squeeze(data.brir_both_modified_Exp5(ids,:,1,reverbLevel,:)); % with given direct sound part
target_dir = squeeze(data.brir_both_modified_Exp5(ids,:,3,reverbLevel,:));
[E_predict_exp5(ids,3,k)] = li2020(target_dir,template_dir,target,template,'stim',stimulus.noise_input,'fs',fs,'fsstim',fs, ...
'flow',200, 'fhigh',16000, 'space',1);
% [E_predict_exp5(ids,3,k)] = Li2020([],[],target,template,'stim',stimulus.noise_input,'fs',fs,'fsstim',fs, ...
% 'flow',200, 'fhigh',16000, 'space',1); % option
end
% condition 4: B = 16
for k =1:reverbLevel
template = squeeze(data.brir_both_modified_Exp5(ids,:,1,1,:)); % with given BRIRs
target = squeeze(data.brir_both_modified_Exp5(ids,:,4,k,:));
template_dir = squeeze(data.brir_both_modified_Exp5(ids,:,1,reverbLevel,:)); % with given direct sound part
target_dir = squeeze(data.brir_both_modified_Exp5(ids,:,4,reverbLevel,:));
[E_predict_exp5(ids,4,k)] = li2020(target_dir,template_dir,target,template,'stim',stimulus.noise_input,'fs',fs,'fsstim',fs, ...
'flow',200, 'fhigh',16000, 'space',1);
% [E_predict_exp5(ids,4,k)] = Li2020([],[],target,template,'stim',stimulus.noise_input,'fs',fs,'fsstim',fs, ...
% 'flow',200, 'fhigh',16000, 'space',1); % option
end
% condition 5: B = 64
for k =1:reverbLevel
template = squeeze(data.brir_both_modified_Exp5(ids,:,1,1,:)); % with given BRIRs
target = squeeze(data.brir_both_modified_Exp5(ids,:,5,k,:));
template_dir = squeeze(data.brir_both_modified_Exp5(ids,:,1,reverbLevel,:)); % with given direct sound part
target_dir = squeeze(data.brir_both_modified_Exp5(ids,:,5,reverbLevel,:));
[E_predict_exp5(ids,5,k)] = li2020(target_dir,template_dir,target,template,'stim',stimulus.noise_input,'fs',fs,'fsstim',fs, ...
'flow',200, 'fhigh',16000, 'space',1);
% [E_predict_exp5(ids,5,k)] = Li2020([],[],target,template,'stim',stimulus.noise_input,'fs',fs,'fsstim',fs, ...
% 'flow',200, 'fhigh',16000, 'space',1); % option
end
end
CI95 =[bsxfun(@times, std(squeeze(E_predict_exp5(:,1,:)))/sqrt(length(E_predict_exp5(:,1,:))),...
tinv([0.025 0.975], length(E_predict_exp5(:,1,:))-1)');
bsxfun(@times, std(squeeze(E_predict_exp5(:,2,:)))/sqrt(length(E_predict_exp5(:,2,:))),...
tinv([0.025 0.975], length(E_predict_exp5(:,2,:))-1)');
bsxfun(@times, std(squeeze(E_predict_exp5(:,3,:)))/sqrt(length(E_predict_exp5(:,3,:))),...
tinv([0.025 0.975], length(E_predict_exp5(:,3,:))-1)');
bsxfun(@times, std(squeeze(E_predict_exp5(:,4,:)))/sqrt(length(E_predict_exp5(:,4,:))),...
tinv([0.025 0.975], length(E_predict_exp5(:,4,:))-1)');
bsxfun(@times, std(squeeze(E_predict_exp5(:,5,:)))/sqrt(length(E_predict_exp5(:,5,:))),...
tinv([0.025 0.975], length(E_predict_exp5(:,5,:))-1)')];
CI95 = CI95(2:2:10,:);
data_li2020('exp5','plot');
x_conditions = 1:5;
legend off
errorbar(x_conditions-0.1, mean(squeeze(E_predict_exp5(:,1,:))), CI95(1,:),'r', 'LineStyle','none','LineWidth',1.5); hold on,
errorbar(x_conditions-0.05, mean(squeeze(E_predict_exp5(:,2,:))), CI95(2,:),'r', 'LineStyle','none','LineWidth',1.5);
errorbar(x_conditions, mean(squeeze(E_predict_exp5(:,3,:))), CI95(3,:),'r', 'LineStyle','none','LineWidth',1.5);
errorbar(x_conditions+0.05, mean(squeeze(E_predict_exp5(:,4,:))), CI95(4,:),'r', 'LineStyle','none','LineWidth',1.5);
errorbar(x_conditions+0.1, mean(squeeze(E_predict_exp5(:,5,:))), CI95(5,:),'r', 'LineStyle','none','LineWidth',1.5);
plot(x_conditions-0.1, median(squeeze(E_predict_exp5(:,1,:))),'ro--', 'LineWidth',1.5, 'MarkerSize',10, 'MarkerFaceColor','r'); hold on,
plot(x_conditions-0.05, median(squeeze(E_predict_exp5(:,2,:))),'rs--', 'LineWidth',1.5, 'MarkerSize',10, 'MarkerFaceColor','r');
plot(x_conditions, median(squeeze(E_predict_exp5(:,3,:))),'rd--', 'LineWidth',1.5, 'MarkerSize',10, 'MarkerFaceColor','r');
plot(x_conditions+0.05, median(squeeze(E_predict_exp5(:,4,:))),'rv--', 'LineWidth',1.5, 'MarkerSize',10, 'MarkerFaceColor','r');
plot(x_conditions+0.1, median(squeeze(E_predict_exp5(:,5,:))),'r^--', 'LineWidth',1.5, 'MarkerSize',10, 'MarkerFaceColor','r');
end
end
Build with Bootstrap