%% INTRODUCTION % the following scripts provide the basic idea of the usage of [nsltools] % you are welcomed to report any bugs, type-errors as well as suggestions to % [email protected] (Taishih Chi), [email protected] (Shihab Shamma)

%% SETUP % make sure the nsltools is in the path list % load colormap, filterbank and parameters loadload;

% you may change the parameters, see WAV2AUD % paras(1): frame jump, in ms, e.g., 4, 8, 16, 32 ms % paras(2): Time constant, in ms, e.q., 4, 8, 16, 32, 64 ms % paras(3): Nonlinear factor, e.g., .1 % paras(4): Octave shift, e.g., -1 for 8k, 0 for 16k (sampling rate) % rv: rate vector, e.g., 2.^(1:5) or 2.^(1:.25:5); (Hz) % sv: scale vector, e.g., 2.^(-2:3) or 2.^(-2:.25:3); (cycle/octave)

%% WAVFORM % sequential format: see LOADFILE % .au format: see AUREAD % .aiff format: see AIFFREAD %x = loadfile('_come.sho'); x = auread('_come.au'); % 8k sample rate

% down sample to 8 kHz if necessary % if the sample rate is not the power of 2, you need to use RESAMPLE % change the paras(4) according to the sample rate %x = decimate(x, 2); %x = x(2:2:length(x));

% if you want to see a schematic plot, you may run xh = schematc(x);

% make the sequence as unit sequence, i.e., ~ N(0, 1) or % zero-mean and unit variance (optional but recommended) % besides, the length 8192, 16384, 32768, etc are preferrable x = unitseq(x);

% play the sound if you like soundsc(x, 8000);

%% SPECTRUM % y is an M-by-128 matrix, M is the # of frames % a short message will be displayed for every complete octave y = wav2aud(x, paras);

% plot the spectrogram aud_plot(y, paras);

%% CORTICAL REPRESENTATION % static representation (e.g., 60th frame of the spectrogram) z = aud2cors(y(60, :), sv); % plot it cor_plts(z, sv, paras);

% dynamic 4-D representation % you can assign an output filename for saving the data see AUD2COR % The process will take a while. 15 seconds for 1 second input data on SPARC Ultra-5_10 machine with default parameters fcorname = '_come.cor'; cr = aud2cor(y, paras, rv, sv, fcorname);

% show the rate-scale energy plot as a function of time % compute rate-scale-time energy matrix from cortical file rst = cor2rst(fcorname);

% or computer rate-scale-time energy matrix directly from 4-D variable cr rst = mean(abs(cr), 4);

% change back to the jet colormap for displaying purpose colormap(jet); rst_view(rst, rv, sv);

%% MANIPULATIONS % at this point, you may do some manipulations % read Contents.m for Manipulations section

%% SPECTROGRAM RECONSTRUCTION yh = cor2aud(fcorname); yh = aud_fix(yh);

%% WAVFORM RECONSTRUCTION xh = aud2wav(yh, [], [paras 10 1 1]);