BAUMGARTNER2014 - Model for localization in sagittal planes

Usage

[p,rang] = baumgartner2014( target,template )
[p,rang,tang] = baumgartner2014( target,template )
[p,rang,tang] = baumgartner2014( target,template,fs,S,lat,stim,fsstim )
[err,pred] = baumgartner2014( target,template,errorflag )

Input parameters

`target`	binaural impulse response(s) referring to the directional transfer function(s) (DFTs) of the target sound(s). Option 1: given in SOFA format -> sagittal plane DTFs will be extracted internally. Option 2: binaural impulse responses of all available listener-specific DTFs of the sagittal plane formated according to the following matrix dimensions: time x direction x channel/ear
`template`	binaural impulse responses of all available listener-specific DTFs of the sagittal plane referring to the perceived lateral angle of the target sound. Options 1 & 2 equivalent to target.

Output parameters

`p`	predicted probability mass vectors for response angles with respect to target positions 1st dim: response angle 2nd dim: target angle
`rang`	polar response angles (after regularization of angular sampling)
`tang`	polar target angles (usefull if sagittal-plane HRTFs are extracted directly from SOFA object)
`err`	predicted localization error (acc. to performance measure defined in errorflag
`pred`	structure with fields p, rang, tang

Description

baumgartner2014(...) is a model for sound-source localization in sagittal planes (SPs). It bases on the comparison of internal sound representation with a template and results in a probabilistic prediction of polar angle response.

baumgartner2014 accepts the following optional parameters:

`'fs',fs`	Define the sampling rate of the impulse responses. Default value is 48000 Hz.
`'S',S`	Set the listener-specific sensitivity threshold (threshold of the sigmoid link function representing the psychometric link between transformation from the distance metric and similarity index) to S. Default value is 1.
`'gamma',G`	Set the degree of selectivity (slope of the sigmoid link function representing the psychometric link between transformation from the distance metric and similarity index) to G. Default value is 6.
`'lat',lat`	Set the apparent lateral angle of the target sound to lat. Default value is 0 degree (median SP).
`'stim',stim`	Define the stimulus (source signal without directional features). As default an impulse is used.
`'fsstim',fss`	Define the sampling rate of the stimulus. Default value is 48000 Hz.
`'flow',flow`	Set the lowest frequency in the filterbank to flow. Default value is 700 Hz.
`'fhigh',fhigh`	Set the highest frequency in the filterbank to fhigh. Default value is 18000 Hz.
`'space',sp`	Set spacing of auditory filter bands (i.e., distance between neighbouring bands) to sp in number of equivalent rectangular bandwidths (ERBs). Default value is 1 ERB.
`'do',do`	Set the differential order of the spectral gradient extraction to do. Default value is 1 and includes restriction to positive gradients inspired by cat DCN functionality.
`'bwcoef',bwc`	Set the binaural weighting coefficient bwc. Default value is 13 degrees.
`'polsamp',ps`	Define the polar-angle sampling of the acoustic data provided for the current sagittal plane. As default the sampling of ARI's HRTFs in the median SP is used, i.e., ps = [-30:5:70,80,100,110:5:210] degrees.
`'rangsamp',rs`	Define the equi-polar sampling of the response predictions. The default is rs = 5 degrees.
`'mrsmsp',eps`	Set the motoric response scatter eps within the median sagittal plane. Default value is 17 degrees.

baumgartner2014 accepts the following flags:

`'regular'`	Apply spline interpolation in order to regularize the angular sampling of the polar response angle. This is the default.
`'noregular'`	Disable regularization of angular sampling.
`'errorflag'`	May be one of the error flags defined in `baumgartner2014pmv2ppp` or `localizationerror`.

Requirements:

SOFA API from http://sourceforge.net/projects/sofacoustics for Matlab (in e.g. thirdparty/SOFA)
Data in hrtf/baumgartner2014
Circular Statistics Toolbox from http://www.mathworks.com/matlabcentral/fileexchange/10676-circular-statistics-toolbox--directional-statistics-

References:

R. Baumgartner, P. Majdak, and B. Laback. Modeling sound-source localization in sagittal planes for human listeners. The Journal of the Acoustical Society of America, 136(2):791-802, 2014. [ DOI ]

R. Lyon. All pole models of auditory filtering. In E. R. Lewis, G. R. Long, R. F. Lyon, P. M. Narins, C. R. Steele, and E. Hecht-Poinar, editors, Diversity in auditory mechanics, pages 205-211. World Scientific Publishing, Singapore, 1997.