THE AUDITORY MODELING TOOLBOX

Applies to version: 1.0.0

View the code

Go to function

ZIEGELWANGER2014 - Direction-continuous model of time-of-arrival (TOA) in HRTFs (robust)

Usage

[Obj,results]=ziegelwanger2014(Obj, data,estimation,outlierDetection,model,p0_onaxis)

Input parameters

Obj SOFA object

Output parameters

Obj SOFA Object
results struct containing the estimated time of arrival

Description

Estimates the Time-of-Arrival for each measurement in Obj (SOFA) and corrects the results with a geometrical model of the head.

The results struct contains the following fields:

'.toa data matrix with time of arrival (TOA) for each impulse response (IR)'

'.p_onaxis estimated on-axis model-parameters'

'.p_offaxis estimated off-axis model-parameters'

Optional input parameters:

'estimation' TOA estimation method, pre-estimated TOAs
'model' correct estimated toa, using geometrical TOA-Model (0: TOA estimated, 1: off-axis TOA modeled (default), 2: on-axis TOA modeled
'p0_onaxis' startvalues for lsqcurvefit
'lowpass' bandwidth setting when used with the estimator from 'itdestimator ('lp for lowpass, 'bb for broadband)
'upper_cutfreq' lowpass cutoff (Hz) when used with the estimator from 'itdestimator'
'threshlvl' threshold level (dB) when used with the 'Threshold estimator from 'itdestimator'
'outlierDetection' detect outliers in estimated TOAs (0...off, 1...on, default values...[0.05; 0.01]) [alpha r] reject outliers using the extreme Studentized deviance test with the significance level of ALPHA and upper bound of outlier rate R

'ziegelwanger2014' accepts the following TOA estimation methods:

'1',toaest Built-in threshold detection
'2',toaest Built-in centroid of squared IR
'3',toaest Built-in mean group delay
'4',toaest Built-in maximum of the minimum-phase cross-correlation (default)
'String',toaest An estimator from the 'itdestimator' function ('MaxIACCe' recommended)

Requirements:

  1. SOFA API from http://sourceforge.net/projects/sofacoustics for Matlab (in e.g. thirdparty/SOFA)
  2. Optimization Toolbox for Matlab
  3. Data in hrtf/ziegelwanger2014

Examples:

To calculate the model parameters for the on-axis time-of-arrival model (p_onaxis) and for the off-axis time-of-arrival model (p_offaxis) for a given HRTF set (SOFA object, 'Obj') with the minimum-phase cross-correlation estimation, use:

[Obj,results]=ziegelwanger2014(Obj,4,1);

References:

H. Ziegelwanger and P. Majdak. Modeling the direction-continuous time-of-arrival in head-related transfer functions. J. Acoust. Soc. Am., 135:1278--1293, 2014.