THE AUDITORY MODELING TOOLBOX
Go to function
hauth2020_sii - calculates the SII according to ANSI S3.5-1997
Program code:
function [sii,A,Z] = hauth2020_sii(E,N,T,I,G)
%hauth2020_sii calculates the SII according to ANSI S3.5-1997
%
% Usage:
% sii = hauth2020_sii(E,N,T,I,G)
%
% Input parameter:
% E : Speech Spectrum Level (Section 3.6 in the standard)
% N : Equivalent Noise Spectrum Level (Section 3.15 in the standard)
% T : Equivalent Hearing Threshold Level [dBHL] (Section 3.23 in the standard)
% I : Band Importance function (Section 3.1 in the standard)
% G : Insertion Gain [dB] (Section 3.28 in the standard)
%
% Output parameter:
% sii : Speech Intelligibility Index
%
% modified by rainerb on June 19, 2006
% modified by rainerb on May 22, 2007
% comments added by rns on Nov 18, 2008
%
% Url: http://amtoolbox.sourceforge.net/amt-0.10.0/doc/modelstages/hauth2020_sii.php
% Copyright (C) 2009-2020 Piotr Majdak and the AMT team.
% This file is part of Auditory Modeling Toolbox (AMT) version 1.0.0
%
% 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/>.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%% 1. handling of input arguments, setting of general parameters
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
if nargin < 5
help(mfilename);
return;
end
if T == 0
T = zeros(size(E)); % No hearing loss
elseif length(T) ~= length(E)
error('Equivalent Hearing Threshold Level: Vector size incorrect');
end % if T == 0
if G == 0
G = zeros(size(E)); % No insertion gain
elseif length(G) ~= length(E)
error('Insertion Gain: Vector size incorrect');
end % if G == 0
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%% 2. get frequency bands and variables according to number of frequency bands
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
switch length(E)
case 6
freqb.name = 'octave'; %octave frequency bands
freqb = local_frequencybands(freqb,I);
case 17
freqb.name = 'equal_cb'; % equal-contributing critical frequency bands
freqb = local_frequencybands(freqb,I);
case 18
freqb.name = 'one_third_octave'; % one-third octave frequency bands
freqb = local_frequencybands(freqb,I);
case 21
freqb.name = 'critical_band'; % critical frequency bands
freqb = local_frequencybands(freqb,I);
case 30
freqb.name = 'gammatone'; % gammatone
freqb = local_frequencybands(freqb,I);
G = 0; % don't use insertion gain
otherwise
freqb.name = 'critical_band';
freqb = local_frequencybands(freqb,I);
c_bw = freqb.bands(:,2)-freqb.bands(:,1);
c_f0 = freqb.bands(:,3);
c_th = freqb.bands(:,4);
c_sp = freqb.bands(:,5);
c_im = freqb.importance;
clear freqb;
freqb.name = 'gammatone';
g_f0 = G(:,1);
g_bw = G(:,2);
freqb.importance = interp1(c_f0,c_im./c_bw,g_f0,'linear','extrap').*g_bw;
freqb.importance(freqb.importance<0) = 0;
freqb.importance = freqb.importance/sum(freqb.importance);
freqb.bands(:,3) = g_f0(:);
freqb.bands(:,1) = -g_bw/2+sqrt(g_bw.^2/4+g_f0.^2);
freqb.bands(:,2) = +g_bw/2+sqrt(g_bw.^2/4+g_f0.^2);
freqb.bands(:,4) = interp1(c_f0,c_th,g_f0,'linear','extrap');
freqb.bands(:,5) = interp1(c_f0,c_sp,g_f0,'linear','extrap');
G = 0;
%error('Equivalent speech Spectrum level: Vector size must be 6,17,18 or 21 (or 30)');
end % switch length(E)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%% 3. calculate SII according to ANSI S3.5-1997
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Equivalent Speech Spectrum Level (5.1.3, Eq. 17):
% speech spectrum level + insertion gain
E = E(:) + G(:);
% Self-Speech Masking Spectrum (4.3.2.1 Eq. 5)
V = E - 24;
% maximum of equivalent noise spectrum level and self-speech masking
% spectrum level (4.3.2.2)
B = max(V,N(:)+G(:));
% Calculate slope parameter Ci for spread of masking (4.3.2.3 )
switch freqb.name
case 'octave'
case 'equal_cb'
case 'one_third_octave'
% Eq. 7
C = 0.6 .* (B + 10 * log10(freqb.bands(:,2)) - 6.353) - 80;
case 'critical_band'
% Eq. 6
C = 0.6.* (B + 10 * log10(freqb.bands(:,2) - freqb.bands(:,1))) - 80;
case 'gammatone'
C = 0.6.* (B + 10 * log10(freqb.bands(:,2) - freqb.bands(:,1))) - 80;
end % switch freqb.
% Initialize Equivalent Masking Spectrum Level (4.3.2.4)
% (mod by rainerb, 2007-05-22)
Z = B(:);
% Calculate Equivalent Masking Spectrum Level (4.3.2.5)
switch freqb.name
case 'octave'
case 'equal_cb'
case 'one_third_octave'
% Eq. 9
for i = 2:18 % for lowest band no effect of upwards spread of masking
Z(i) = 10 * log10(10.^(0.1 * N(i)) + sum(10.^(0.1 * (B(1:(i-1)) + 3.32 .* ...
C(1:(i-1)) .* log10(0.89 ...
* freqb.bands(i,3) ./ freqb.bands(1:(i-1),3))))));
end;
case 'critical_band'
% Eq. 8
for i = 2:21 % for lowest band no effect of upwards spread of masking
Z(i) = 10 * log10(10.^(0.1 * N(i)) + sum(10.^(0.1 * (B(1:(i-1)) + 3.32 .* ...
C(1:(i-1)) .* ...
log10(freqb.bands(i,3) ./ ...
freqb.bands(1:(i-1),2))))));
end;
case 'gammatone'
%% rainerb, 2008-06-19: spread of masking is already included in gammatone
end % switch freqb.
% Equivalent Internal Noise Spectrum Level (4.4 Eq. 10):
% reference internal noise spectrum level (4th column of freqb.bands) + dB HL
X = freqb.bands(:,4) + T(:);
% Disturbance Spectrum Level (4.5):
% maximum of equivalent internal noise spectrum level and eqivalent
% masking spectrum level
D = max(Z,X);
% Level Distortion Factor (4.6 Eq. 11)
L = 1 - (E - freqb.bands(:,5) - 10) ./ 160;
L = ((L <= 1) .* L ) + (L > 1);
L = 1; % How does distortion fit to using external noise for simulating the hearing threshold in BSIM?
% Band Audibility Function
% temporary variable K (4.7.1 Eq. 12)
K = (E - D + 15) / 30;
K = K .* (K > 0); % limit K to [0,1]
K = ((K <= 1) .*K ) + (K > 1); % limit K to [0,1]
% band audibility function (7.7.2 Eq. 13):
% product of distortion factor and temporal variable
A = L .* K;
% Speech Intelligibility Index (4.8 Eq. 14)
S = sum(freqb.importance .* A);
sii = S;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%/////////////////////////////////////////////////////////////////////////////////%
%% private functions
%/////////////////////////////////////////////////////////////////////////////////%
function freqb = local_frequencybands(freqb,I)
%this function selects the frequency bands, center frequencies etc. needed for calculation
switch freqb.name
case 'octave'
freqb.bands = ...
[
177 354 250 -3.9 34.75; ... % center freq., reference internal noise spectrum level, standard speech spectrum level for normal speech
353 707 500 -9.7 34.27; ...
707 1414 1000 -12.5 25.01; ...
1414 2828 2000 -17.7 17.32; ...
2828 5657 4000 -25.9 9.33; ...
5657 11314 8000 -7.1 1.13];
freqb.importance = [0.06 0.17 0.24 0.26 0.21 0.05]';
case 'equal_cb'
freqb.bands = [];
freqb.importance = [];
case 'one_third_octave'
freqb.bands = ...
[
143 179 160 0.60 32.41
179 224 200 -1.70 34.48
224 281 250 -3.90 34.75
281 355 315 -6.10 33.98
355 447 400 -8.20 34.59
447 561 500 -9.70 34.27
561 710 630 -10.80 32.06
710 894 800 -11.90 28.30
894 1118 1000 -12.50 25.01
1118 1414 1250 -13.50 23.00
1414 1789 1600 -15.40 20.15
1789 2236 2000 -17.70 17.32
2236 2806 2500 -21.20 13.18
2806 3550 3150 -24.20 11.55
3550 4472 4000 -25.90 9.33
4472 5612 5000 -23.60 5.31
5612 7099 6300 -15.80 2.59
7099 8980 8000 -7.10 1.13
];
% Band importance functions:
% 1: standard of Table 3
% 2: various nonsense syllable tests where most English phonemes occur equally often
% 3: CID-22
% 4: NU6
% 5: Diagnostic Rhyme test
% 6: short passages of easy reading material
% 7: SPIN
% 8: Easy speech (Pavlovic 87, JASA 82,413-422, Table II)
freqb.importance = ...
[
% 0.00083 0 0.0365 0.0168 0 0.0114 0 0.0114
% 0.0095 0 0.0279 0.013 0.024 0.0153 0.0255 0.0153
% 0.015 0.0153 0.0405 0.0211 0.033 0.0179 0.0256 0.0179
% 0.0289 0.0284 0.05 0.0344 0.039 0.0558 0.036 0.0558
% 0.044 0.0363 0.053 0.0517 0.0571 0.0898 0.0362 0.0898
% 0.0578 0.0422 0.0518 0.0737 0.0691 0.0944 0.0514 0.0944
% 0.0653 0.0506 0.0514 0.0658 0.0781 0.0709 0.0616 0.0709
% 0.0711 0.0584 0.0575 0.0644 0.0751 0.066 0.077 0.0660
% 0.0818 0.0667 0.0717 0.0664 0.0781 0.0628 0.0718 0.0628
% 0.0844 0.0774 0.0873 0.0802 0.0811 0.0672 0.0718 0.0672
% 0.0882 0.0893 0.0902 0.0987 0.0961 0.0747 0.1075 0.0747
% 0.0898 0.1104 0.0938 0.1171 0.0901 0.0755 0.0921 0.0755
% 0.0868 0.112 0.0928 0.0932 0.0781 0.082 0.1026 0.0820
% 0.0844 0.0981 0.0678 0.0783 0.0691 0.0808 0.0922 0.0808
% 0.0771 0.0867 0.0498 0.0562 0.048 0.0483 0.0719 0.0483
% 0.0527 0.0728 0.0312 0.0337 0.033 0.0453 0.0461 0.0453
% 0.0364 0.0551 0.0215 0.0177 0.027 0.0274 0.0306 0.0274
% 0.0185 0 0.0253 0.0176 0.024 0.0145 0 0.0145
0.0008 0.0000 0.0365 0.0168 0.0000 0.0114 0.0000 0.0114 0.0000
0.0095 0.0000 0.0279 0.0130 0.0240 0.0153 0.0255 0.0153 0.0000
0.0150 0.0153 0.0405 0.0211 0.0330 0.0179 0.0256 0.0179 0.0000
0.0289 0.0284 0.0500 0.0344 0.0390 0.0558 0.0360 0.0558 0.0000
0.0440 0.0363 0.0530 0.0517 0.0571 0.0898 0.0362 0.0898 0.0000
0.0578 0.0422 0.0518 0.0737 0.0691 0.0944 0.0514 0.0944 1.0000
0.0653 0.0506 0.0514 0.0658 0.0781 0.0709 0.0616 0.0709 0.0000
0.0711 0.0584 0.0575 0.0644 0.0751 0.0660 0.0770 0.0660 0.0000
0.0818 0.0667 0.0717 0.0664 0.0781 0.0628 0.0718 0.0628 0.0000
0.0844 0.0774 0.0873 0.0802 0.0811 0.0672 0.0718 0.0672 0.0000
0.0882 0.0893 0.0902 0.0987 0.0961 0.0747 0.1075 0.0747 0.0000
0.0898 0.1104 0.0938 0.1171 0.0901 0.0755 0.0921 0.0755 0.0000
0.0868 0.1120 0.0928 0.0932 0.0781 0.0820 0.1026 0.0820 0.0000
0.0844 0.0981 0.0678 0.0783 0.0691 0.0808 0.0922 0.0808 0.0000
0.0771 0.0867 0.0498 0.0562 0.0480 0.0483 0.0719 0.0483 0.0000
0.0527 0.0728 0.0312 0.0337 0.0330 0.0453 0.0461 0.0453 0.0000
0.0364 0.0551 0.0215 0.0177 0.0270 0.0274 0.0306 0.0274 0.0000
0.0185 0.0000 0.0253 0.0176 0.0240 0.0145 0.0000 0.0145 0.0000
];
if 1 == max(size(I))
freqb.importance = freqb.importance(:,I);
else
if 18 == length(I(:))
freqb.importance = I(:)/sum(I(:));
else
error('wrong band importance function');
end
end
case 'critical_band'
freqb.bands = ...
[
100 200 150 1.50 31.44; % lower and upper band limit, center freq
200 300 250 -3.90 34.75; % [Hz], internal noise spectrum
300 400 350 -7.20 34.14; % level [dB] and Standard
400 510 450 -8.90 34.58; % speech spectrum level
510 630 570 -10.30 33.17; % at normal vocal effort
630 770 700 -11.40 30.34;
770 920 840 -12.00 27.59;
920 1080 1000 -12.50 25.01;
1080 1270 1170 -13.20 23.52;
1270 1480 1370 -14.00 22.28;
1480 1720 1600 -15.40 20.15;
1720 2000 1850 -16.90 18.29;
2000 2320 2150 -18.80 16.37;
2320 2700 2500 -21.20 13.80;
2700 3150 2900 -23.20 12.21;
3150 3700 3400 -24.90 11.09;
3700 4400 4000 -25.90 9.33;
4400 5300 4800 -24.20 5.84;
5300 6400 5800 -19.00 3.47;
6400 7700 7000 -11.70 1.78;
7700 9500 8500 -6.00 -0.14
];
freqb.importance = ...
[
0.0130; % Band importance for SPIN
0.0478;
0.0451;
0.0470;
0.0523;
0.0591;
0.0591;
0.0503;
0.0503;
0.0556;
0.0699;
0.0625;
0.0602;
0.0684;
0.0638;
0.0605;
0.0534;
0.0394;
0.0291;
0.0132;
0.0000
];
case 'gammatone'
freqb.bands = ...
[
126.33 166.81 146.02 1.7149 31.308 % 18.778 30
166.8 211.89 188.74 -0.59196 32.722 % 14.657 31
211.9 262.13 236.34 -3.1624 34.298 % 11.52 33.66
262.14 318.09 289.36 -5.1989 34.51 % 9.065 34.341
318.09 380.43 348.42 -7.1479 34.15 % 7.0957 34.337
380.43 449.87 414.21 -8.2916 34.423 % 5.43 34.371
449.87 527.22 487.5 -9.3375 34.139 % 4.0375 34.4
527.23 613.4 569.15 -10.29 33.18 % 3.0553 33.49
613.4 709.39 660.1 -11.062 31.209 % 2.2938 31.978
709.39 816.32 761.41 -11.663 29.134 % 1.9362 29.834
816.32 935.44 874.27 -12.107 27.037 % 1.9485 27.753
935.44 1068.1 1000 -12.5 25.01 % 2.2 25.655
1068.1 1216 1140.1 -13.077 23.782 % 2.5922 24.211
1216 1380.6 1296.1 -13.704 22.738 % 2.7157 22.995
1380.6 1564.1 1469.9 -14.608 21.355 % 2.0205 21.775
1564.1 1768.4 1663.5 -15.781 19.678 % 1.0715 20.176
1768.4 1996 1879.2 -17.085 18.103 % -0.38433 18.634
1996 2249.6 2119.4 -18.606 16.566 % -1.9165 16.952
2249.6 2532.1 2387.1 -20.426 14.629 % -3.5223 15.016
2532.1 2846.8 2685.2 -22.126 13.064 % -4.7983 13.308
2846.8 3197.3 3017.3 -23.599 11.947 % -5.8713 12.281
3197.3 3587.8 3387.3 -24.857 11.118 % -6.2442 11.329
3587.8 4022.8 3799.4 -25.566 9.9184 % -6.1472 10.347
4022.8 4507.4 4258.6 -25.35 8.2019 % -5.1692 8.9881
4507.4 5047.2 4770 -24.264 5.9709 % -3.328 7.065
5047.2 5648.5 5339.7 -21.394 4.5609 % -0.40942 5.1768
5648.5 6318.4 5974.4 -17.939 3.2244 % 3.4962 3.8291
6318.4 7064.6 6681.4 -13.638 2.2287 % 7.2948 2.6833
7064.6 7895.8 7469 -9.9178 1.1797 % 11.001 1.8332
7895.8 8821.8 8346.3 -6.5841 0.056736 % 13.725 0
];
freqb.importance = ...
[
0.0076 0.0047 0.0050
0.0108 0.0120 0.0655
0.0149 0.0217 0.0225
0.0311 0.0263 0.0035
0.0573 0.0283 0.0596
0.0650 0.0304 0.0761
0.0649 0.0334 0.0159
0.0532 0.0377 0.0023
0.0473 0.0411 0.0000
0.0418 0.0440 0.0743
0.0380 0.0451 0.0617
0.0366 0.0419 0.0062
0.0355 0.0406 0.0178
0.0385 0.0438 0.0004
0.0405 0.0509 0.0058
0.0402 0.0562 0.0435
0.0397 0.0502 0.0153
0.0396 0.0488 0.0394
0.0416 0.0518 0.0827
0.0424 0.0513 0.0485
0.0406 0.0473 0.0209
0.0355 0.0434 0.0013
0.0280 0.0382 0.0305
0.0236 0.0320 0.0259
0.0235 0.0244 0.0342
0.0195 0.0208 0.0183
0.0150 0.0162 0.0425
0.0139 0.0108 0.0658
0.0103 0.0058 0.0172
0.0036 0.0010 0.0347
];
if 1 == max(size(I))
freqb.importance = freqb.importance(:,I);
else
if 30 == length(I(:))
freqb.importance = I(:)/sum(I(:));
else
error('wrong band importance function');
end
end
end % switch fb
return
Build with Bootstrap