THE AUDITORY MODELING TOOLBOX

Applies to version: 0.9.9

View the code

Go to function

lopezpoveda2001 - Dual Resonance Nonlinear Filterbank

Usage

outsig = lopezpoveda2001(insig,fs);

lopezpoveda2001(insig,fs) computes the Dual Resonance Non-Linear (lopezpoveda2001) filterbank of the input signal insig sampled at fs Hz with channels specified by the center frequencies in fc. The lopezpoveda2001 is described in the paper Lopez-Poveda and Meddis (2001). The lopezpoveda2001 models the basilar membrane non-linearity.

This version of the lopezpoveda2001 incoorperate the middle-ear filter used in Lopez-Poveda and Meddis (2001).

The lopezpoveda2001 takes a lot of parameters which vary over frequency. Such a parameter is described by a \(1 \times 2\) vector [b a] and indicates that the value of the parameter at frequency fc can be calculated by

\begin{equation*} 10^{b+a\cdot log_{10}(fc)} \end{equation*}

The parameters are:

'flow',flow Set the lowest frequency in the filterbank to flow. Default value is 80 Hz.
'fhigh',fhigh Set the highest frequency in the filterbank to fhigh. Default value is 8000 Hz.
'basef',basef Ensure that the frequency basef is a centre frequency in the filterbank. The default value of [] means no default.
'middleear' Perform middleear filtering before the actual lopezpoveda2001 is applied using the middleear filter specified in Lopez-Poveda and Meddis (2001), and compensate for the effect of the filter after lopezpoveda2001 filtering. This is the default.
'nomiddleear' No middle-ear filtering. Be carefull with this setting, as another scaling must then be perform to convert the input to stapes movement.
'bothparts' Compute both the linear and the non-linear path of the lopezpoveda2001. This is the default.
'linonly' Compute only the linear path.
'nlinonly' Compute only the non-linear path.
'lin_ngt',n Number of cascaded gammatone filter in the linear part, default value is 2.
'lin_nlp',n Number of cascaded lowpass filters in the linear part, default value is 4
'lin_gain',g Gain in the linear part, default value is [4.20405 ... .47909].
'lin_fc',fc Centre frequencies of the gammatone filters in the linear part. Default value is [-0.06762 1.01679].
'lin_bw',bw Bandwidth of the gammatone filters in the linear part. Default value is [.03728 .78563]
'lin_lp_cutoff',c Cutoff frequency of the lowpass filters in the linear part. Default value is [-0.06762 1.01679 ]
'nlin_ngt_before',n Number of cascaded gammatone filters in the non-linear part before the broken stick non-linearity. Default value is 3.
'nlin_ngt_after',n Number of cascaded gammatone filters in the non-linear part after the broken stick non-linearity. The default value of [] means to use the 'before' value.
'nlin_nlp',n Number of cascaded lowpass filters in the non-linear part. Default value is 3.
'nlin_fc_before',fc Center frequencies of the gammatone filters in the non-linear part before the broken stick non-linearity. Default value is [-0.05252 1.01650].
'nlin_fc_after',fc Center frequencies of the gammatone filters in the non-linear part after the broken stick non-linearity. The default value of [] means to use the 'before' value.
'nlin_bw_before',bw Bandwidth of the gammatone filters in the non-linear part before the broken stick non-linearity. Default value is [-0.03193 .77426 ].
'nlin_bw_after',w Bandwidth of the gammatone filters in the non-linear part after the broken stick non-linearity. The default value of [] means to use the 'before' value.
'nlin_lp_cutoff',c Cutoff frequency of the lowpass filters in the non-linear part. Default value is [-0.05252 1.01650 ].
'nlin_a',a The a coefficient for the broken-stick non-linearity. Default value is [1.40298 .81916 ].
'nlin_b',b The b coefficient for the broken-stick non-linearity. Default value is [1.61912 -.81867].
'nlin_c',c The c coefficient for the broken-stick non-linearity. Default value is [-.60206 0].
'nlin_d',d The d coefficient for the broken-stick non-linearity. Default value is 1.

The output from lopezpoveda2001 can be conveniently visualized using the plotfilterbank function from LTFAT.

References:

E. Lopez-Poveda and R. Meddis. A human nonlinear cochlear filterbank. J. Acoust. Soc. Am., 110:3107-3118, 2001.

R. Meddis, L. O'Mard, and E. Lopez-Poveda. A computational algorithm for computing nonlinear auditory frequency selectivity. J. Acoust. Soc. Am., 109:2852-2861, 2001.