This repo includes the files necessary to run the Ambisonic HRTF pre-processing code in Matlab. If you use this code, please cite Thomas McKenzie's PhD thesis (University of York, 2019) available here:
T. McKenzie: "High frequency reproduction in binaural Ambisonic rendering." PhD Thesis, University
of York, 2019.
See \test_scripts folder for test scripts:
Run load_ambisonic_configuration to generate binaural Ambisonic decoders with the pre-processing techniques. Configure this script to select the desired pre-processing techniques. It will produce one pre-processed binaural Ambisonic decoder, and one with no pre-processing (labelled NPP). The included HRTFs are from the Benjamin Bernschutz KU 100 database.
To test the decoders, run test_ambisonic_decoder. This will produce a figure with the diffuse-field responses of both decoders. It will also compare binaural Ambisonic rendering using the two decoders with standard (non-Ambisonic) HRIRs in the folder \hrirs. This comparison is done for predicted binaural colouration, using the mckenzie2022 model as implemented in the Auditory Modelling Toolbox. Please cite the following paper for this:
T. McKenzie, C. Armstrong, L. Ward, D. Murphy, and G. Kearney. Predicting the colouration between
binaural signals. Appl. Sci., 12(2441), 2022.
Other test metrics include change in interaural level difference (ILD) and interaural time difference (ITD).
To listen to the differences between the decoders, run test_sound_percussion, to hear the standard binaural Ambisonic decoder, followed by the pre-processed one, and then an HRTF convolution render. The sound is several percussive sounds virtually panned to various locations (see thesis for exact locations). For another test sound script to hear the differences between the decoders, run test_sound_soundscape to hear the standard binaural Ambisonic decoder followed by the pre-processed one. The soundscape excerpts are from the Marc Green EigenScape database
Thomas McKenzie, University of Edinburgh. Code written whilst at University of York, 2019. thomas.mckenzie@ed.ac.uk
This code requires the Matlab toolboxes Phased Array System Toolbox and Curve Fitting Toolbox to run.
This code uses and includes some functions written by others. In all cases, these are documented in the code when used. Some functions have been adapted. These include:
Ambisonic functions by Archontis Politis: https://uk.mathworks.com/matlabcentral/fileexchange/54833-higher-order-ambisonics-hoa-library,
Inverse filtering code by Matthes: https://uk.mathworks.com/matlabcentral/fileexchange/19294-inverse-fir-filter,
Voronoi sphere code from Bruno Luong: https://uk.mathworks.com/matlabcentral/fileexchange/40989-voronoi-sphere,
Quadrature code from John Burkardt: https://people.sc.fsu.edu/~jburkardt/datasets/sphere_grid/sphere_grid.html,
iso226 code by Jeff Tackett: https://uk.mathworks.com/matlabcentral/fileexchange/7028-iso-226-equal-loudness-level-contour-signal?focused=22bd2900-9b20-b731-8ea3-64189da018b4&tab=function,
ERB code from IOSR toolbox (Surrey University): https://github.com/IoSR-Surrey/MatlabToolbox