comparison toolboxes/MIRtoolbox1.3.2/AuditoryToolbox/LyonPassiveEar.m @ 0:e9a9cd732c1e tip

first hg version after svn

+function y=LyonPassiveEar(x,sr,df,earQ,stepfactor,differ,agcf,taufactor)
+% y=LyonPassiveEar(input, sample_rate, decimation, earQ, stepfactor);
+%
+% Changes
+% Zeroed out the preemphasis channels before doing the AGC.  This is needed
+% so that cochlear inversion is possible (since we don't have those channels,
+% we can't invert their gain.)..... Malcolm 6/26/95
+
+% (c) 1998 Interval Research Corporation  
+
+if nargin < 3
+	fprintf('Syntax: y=LyonPassiveEar(input, sample_rate, decimation, earQ, stepfactor)\n');
+	fprintf(' The input, sample_rate, and decimation parameters are mandatory.\n');
+	return;
+end
+
+if df < 1; df = 1; end
+if nargin < 4; earQ = 8; end
+if nargin < 5; stepfactor = earQ/32; end
+if nargin < 6; differ=1; end
+if nargin < 7; agcf=1; end
+if nargin < 8; taufactor=3; end
+
+earFilters = DesignLyonFilters(sr, earQ, stepfactor);
+
+nSamples = length(x);
+nOutputSamples = floor(nSamples/df);
+[nChannels filterWidth] = size(earFilters);
+
+sosOutput = zeros(nChannels, df);
+sosState = zeros(nChannels, 2);
+agcState = zeros(nChannels, 4);
+y = zeros(nChannels, nOutputSamples);
+
+decEps = EpsilonFromTauFS(df/sr*taufactor,sr);
+decState = zeros(nChannels, 2);
+decFilt = SetGain([0 0 1 -2*(1-decEps) (1-decEps)^2], 1, 0, sr);
+
+eps1 = EpsilonFromTauFS(.64,sr);
+eps2 = EpsilonFromTauFS(.16,sr);
+eps3 = EpsilonFromTauFS(.04,sr);
+eps4 = EpsilonFromTauFS(.01,sr);
+
+tar1 = .0032;
+tar2 = .0016;
+tar3 = .0008;
+tar4 = .0004;
+
+if 0
+	fprintf('df=%g, earq=%g, stepfactor=%g, differ=%g\n',df,earQ,stepfactor,differ);
+	fprintf('agcf=%g, taufactor=%g\n', agcf, taufactor);
+	[tar1 tar2 tar3 tar4; eps1 eps2 eps3 eps4]
+end
+
+for i=0:nOutputSamples-1
+	[sosOutput sosState]= soscascade(x(i*df+1:i*df+df), earFilters, ...  
+				sosState);
+	output = max(0, sosOutput);		%% Half Wave Rectify
+	output(1) = 0;					%% Test Hack to make inversion easier.
+	output(2) = 0;
+	if agcf > 0
+		[output agcState] = agc(output, [tar1 tar2 tar3 tar4; ...
+				eps1 eps2 eps3 eps4], ...
+				agcState);
+	end
+
+	if differ > 0
+		output = [output(1,:);output(1:nChannels-1,:) - ...
+					output(2:nChannels,:)];
+		output = max(0, output);
+	end
+
+	if df > 1
+		[output decState] = sosfilters(output, decFilt, decState);
+	end
+	y(:,i+1) = output(:,df);
+end
+
+%y = min(y,2*tar4);
+y=y(3:nChannels,:);