LINDEMANN1986BINCORR - Cross-correlation between two input signals a la Lindemann

Usage

crosscorr = bincorr(insig,fs,c_s,w_f,M_f,T_int,N_1)

Input parameters

`insig`	signal with \(t \times nfc \times 2\) (right, left channel)
`fs`	sampling rate

Output parameters

`crosscorr`	output matrix containing the correlations
`t`	time axis corresponding to the n time samples in crosscorr

Description

lindemann1986bincorr(insig,fs) is an implementation of the Lindemann cross-correlation algorithm to simulate a binaural delay line. The output is a matrix of size \(n \times m \times amplitude\) , where n = length(t)/fs.

The cross-correlation is calculated using:

\begin{equation*} CC(\tau,t) = \int_{-\infty}^{t} R(l-\frac{\tau}{2}) L(k+\frac{\tau}{2})e^{-(t-k)/T_{int}} dk \end{equation*}

where \(T_{int}\) denotes an integration time constant and R, L the right and left input signal.

lindemann1986bincorr takes the following key/value pairs at the end of the command line:

`'c_s',c_s`	Stationary inhibition factor, \(0 \leq c_s \leq 1\) (0.0 = no inhibition). Default value is 0.3.
`'w_f',w_f`	Monaural sensitivity at the end of the delay line, \(0 \leq w_f < 1\) . Default value is 0.035.
`'M_f',M_f`	Determines the decrease of the monaural sensitivity along the delay line. Default value is 6
`'T_int',t_int`	integration time window (ms). This is the memory of the correlation process with `exp(-1/T_int)`. Also this determines the time steps in the binaural activity map, because every time step T_int a new running cross-correlation is started, so every T_int we have a new result in crosscorr. You can set T_int = inf if you like to have no memory effects, then you will get only one time step in crosscorr. Default value is 5 ms.
`'N_1',N_1`	Sample at which the first running cross-correlation should be started to avoid onset effects (see Lindemann (1986a) p. 1614). Default: 1, 17640 (200/ffs* with \(f=500\) and \(fs=44100\) ) for `'stationary'` (see above)
`'stationary'`	will set the default values of N_1=17640 and T_int=inf, use this for stationary input signals.

The key/values can also be specified first in the line of arguments in the following order: c_s, w_f, M_f, T_int, N_1.

References:

W. Lindemann. Extension of a binaural cross-correlation model by contralateral inhibition. I. Simulation of lateralization for stationary signals. J. Acoust. Soc. Am., 80:1608-1622, 1986.