function plotlindemann1986(crosscorr,t,varargin)
%PLOTLINDEMANN1986 Plots the binaural output pattern of the lindemann model
% Usage: plotlindemann1986(crosscorr,t,f,tstr);
% plotlindemann1986(crosscorr,t,f);
% plotlindemann1986(crosscorr,t,tstr);
% plotlindemann1986(crosscorr,t);
%
% Input parameters:
% crosscorr : cross-correlation matrix, output from the lindemann
% function
% t : time vector of the analysed stimuli (used for t axis)
%
% `plotlindemann1986(crosscorr,t)` plots the cross-correlation output from the
% lindemann function as a so called binaural activity map. This means the
% correlation value is plotted depending on time of the stimulus and
% the correlation-time delay. *t* is the time axis of the plot. *f* determines
% the frequency channel to plot by using the channel in which the
% frequency *f* belongs.
%
% If *crosscorr* has more than one time step a 3D activity map is plotted, else
% a 2D plot of the cross-correlation is done.
%
% The function takes the following flags at the end of the line of
% input arguments:
%
% 'fc',fc plot only the frequency channel with its center frequency
% is nearest to the frequency f. The default value of []
% means to plot the mean about all frequency channels
%
% 'title',t display t as the title overriding the default.
%
% You may also supply the parameters in the input arguments in the
% following order: `plotlindemann1986(crosscorr,t,fc)`
%
% See also: lindemann1986, lindemann1986bincorr
% AUTHOR: Hagen Wierstorf
% ------ Checking of input parameters -----------------------------------
if nargin<2
error('%s: Too few input arguments.',upper(mfilename));
end;
if ~isnumeric(crosscorr)
error('%s: crosscorr has to be numeric!',upper(mfilename));
end
if ( ~isnumeric(t) || ~isvector(t) )
error('%s: t has to be a vector!',upper(mfilename));
end
definput.keyvals.title=[];
definput.keyvals.fc=[];
[flags,keyvals] = ltfatarghelper({'fc','title'},definput,varargin);
if isempty(keyvals.fc)
binpattern = mean(crosscorr,3);
else
% Minimum and maximum frequency in the lindemann model (see lindemann.m)
flow = erbtofreq(5);
fhigh = erbtofreq(40);
if ~isscalar(keyvals.fc)
error('%s: fc has to be a scalar!',upper(mfilename));
elseif keyvals.fc<flow || keyvals.fc>fhigh
error('%s: fc has to be between %.0f Hz and %.0f Hz.',...
upper(mfilename),flow,fhigh);
end
% Calculate the frequency channel to plot
% NOTE: it starts with the fifth channel in the lindemann model, so we have
% to subtract 4 to index the binpattern correctly.
fc = round(freqtoerb(keyvals.fc));
binpattern = crosscorr(:,:,fc-4);
end;
% ------ Computation -----------------------------------------------------
% Calculate tau (delay line time) axes
tau = linspace(-1,1,size(crosscorr,2));
% ------ Plotting --------------------------------------------------------
if size(crosscorr,1)==1
% If we have only one time step (stationary case) plot 2D
plot(tau,binpattern);
else
mesh(tau,t,binpattern);
ylabel('t (s)');
end
xlabel('correlation-time delay (ms)');
% Create title, if fc is given but not tstr
if isempty(keyvals.title) && ~isempty(keyvals.fc)
keyvals.title = sprintf('fc = %i',fc);
end
% Plot title
if ~isempty(keyvals.title)
title(keyvals.title);
end