KELVASA2015_LOCALIZE
uses calibration data to map bilateral spike rate differences to an azimuthal angle

Usage:

[ANbinPredictions, weightedPredictions, mappingData] = ...
kelvasa2015_localize(mappingData, SpkDiffPerBin, SpkSumPerBin, varargin)

Input parameters:

`mappingData`	Structure containing the necessary data to map bilateral spike rate differences to a predicted azimuthal angle.
`SpkDiffPerBin`	N x M matrix of chan2 - chan1 spike rate differences in spikes/sec where N are user defined AN frequency bands and M are time bins over the entire signal.
`SpkSumPerBin`	N x M matrix of chan2 + chan1 spike rate sums in spikes/sec where N are user defined AN frequency bands and M are time bins over the entire signal.

Output parameters:

`ANbinPredictions`	N x M matrix of bin azimuthal angle predictions in degrees with N being the number of user defined AN frequency bands and M being the number of time windows.
`weightedPredictions`
	1 x M vector of bin weighted azimuthal angle predictions in degrees with M being the number of time windows.

Description:

KELVASA2015_localize(APvec,sigLengthSec,varargin)implements a user defined localization model as detailed in (Kelvasa & Dietz (2015)) to map bilateral spike rate differences to a predicted azimuithal source angle.

Fields added to mappingData:

`'RateLevelPerBin'`	N x M matrix of monaural spike rates in spikes/sec with N being the number of user defined AN frequency bands and M being the range of signal levels in dB SPL over which the calibration signal was processed.
`'ILDcentFreq'`	1 x N vector of center frequencies in Hz used in CI processing that correspond to the N user defined AN frequency bands.
`'ILD'`	N x M matrix of interaural level differences in dB SPL for each of the the N user defined AN frequency bands over the M azimuthal angles for which the signal was processed.
`'IldAngSlope'`	1 x N vector of linear regression slopes in dB/degree for each of the the N user defined AN frequency bands.
`'RateLevelSlope'`	1 x N vector of linear regression slopes in spike Rate/dB for each of the the N user defined AN frequency bands.
`'spkDiffAziSlopes'`	1 x N vector of linear regression slopes in bilateral spike rate differences /dB for each of the the N user defined AN frequency bands.
`'Avg'`	M x N vector of spike rate difference averages over all time bins where N are user defined AN frequency bands and M are the range of azimuthal angles over which the signal was processed.
`'covariance'`	M x N vector of spike rate difference covariances over all time bins where N are user defined AN frequency bands and M are the range of azimuthal angles over which the signal was processed.

References:

D. Kelvasa and M. Dietz. Auditory model-based sound direction estimation with bilateral cochlear implants. Trends in Hearing, 19:2331216515616378, 2015. [ DOI ]