Populate database

DirPat_loudspeaker_cube/README.md

@@ -0,0 +1,27 @@
+The data are the measurements of the 4 loudspeaker drivers of the IEM Loudspeaker Cube in the [DirPat database](https://opendata.iem.at/projects/dirpat/). All directivity models are far-field models based on third-octave smoothed magnitude data like in the Common Loudspeaker Format. The details are described in 
+
+> J. Ahrens and S. Bilbao, “Interpolation and Range Extrapolation of Sound Source Directivity Based on a Spherical Wave Propagation Model,” in Proc. of IEEE ICASSP, Barcelona, Spain, May 2020
+
+and particularly in
+
+> J. Ahrens and S. Bilbao, “Computation of Spherical Harmonics Based Sound Source Directivity Models from Sparse Measurement Data,” Forum Acusticum, Lyon, France, May 2020.
+
+
+
+Example balloon plots: 
+
+`irs_DirPat_loudspeaker_cube_1_N6.mat`
+
+![irs_DirPat_loudspeaker_cube_1_N6](irs_DirPat_loudspeaker_cube_1_N6.png "irs_DirPat_loudspeaker_cube_1_N6")
+
+`irs_DirPat_loudspeaker_cube_2_N6.mat`
+
+![irs_DirPat_loudspeaker_cube_2_N6](irs_DirPat_loudspeaker_cube_2_N6.png "irs_DirPat_loudspeaker_cube_2_N6")
+
+`irs_DirPat_loudspeaker_cube_3_N6.mat`
+
+![irs_DirPat_loudspeaker_cube_3_N6](irs_DirPat_loudspeaker_cube_3_N6.png "irs_DirPat_loudspeaker_cube_3_N6")
+
+`irs_DirPat_loudspeaker_cube_4_N6.mat`
+
+![irs_DirPat_loudspeaker_cube_4_N6](irs_DirPat_loudspeaker_cube_4_N6.png "irs_DirPat_loudspeaker_cube_4_N6")

DirPat_singing_voice/README.md

@@ -0,0 +1,54 @@
+The data are the singing voice measurements from the [DirPat database](https://opendata.iem.at/projects/dirpat/), specifically from the files
+
+* `IR_a_long_sweep.mat`
+
+* `IR_a_closed_sweep.mat`
+
+* `IR_a_open_sweep.mat`
+
+* `IR_a_wide_sweep.mat`.
+
+The calibration data in file `cal64.mat` were applied to all datasets. All directivity models are far-field models referenced to the direction straight ahead and computed as described in 
+
+> J. Ahrens and S. Bilbao, “Interpolation and Range Extrapolation of Sound Source Directivity Based on a Spherical Wave Propagation Model,” in Proc. of IEEE ICASSP, Barcelona, Spain, May 2020
+
+and
+
+> J. Ahrens and S. Bilbao, “Computation of Spherical Harmonics Based Sound Source Directivity Models from Sparse Measurement Data,” Forum Acusticum, Lyon, France, May 2020.
+
+
+
+Example balloon plots: 
+
+`irs_DirPat_a_closed_sweep_N9_non-reg.mat`
+
+![irs_DirPat_a_closed_sweep_N9_non-reg](irs_DirPat_a_closed_sweep_N9_non-reg.png "irs_DirPat_a_closed_sweep_N9_non-reg")
+
+`irs_DirPat_a_closed_sweep_N9_reg.mat` 
+
+![irs_DirPat_a_closed_sweep_N9_reg](irs_DirPat_a_closed_sweep_N9_reg.png "irs_DirPat_a_closed_sweep_N9_reg")
+
+`irs_DirPat_a_long_sweep_N9_non-reg.mat`
+
+![irs_DirPat_a_long_sweep_N9_non-reg](irs_DirPat_a_long_sweep_N9_non-reg.png "irs_DirPat_a_long_sweep_N9_non-reg")
+
+`irs_DirPat_a_long_sweep_N9_reg.mat` 
+
+![irs_DirPat_a_long_sweep_N9_reg](irs_DirPat_a_long_sweep_N9_reg.png "irs_DirPat_a_long_sweep_N9_reg")
+
+`irs_DirPat_a_open_sweep_N9_non-reg.mat`
+
+![irs_DirPat_a_open_sweep_N9_non-reg](irs_DirPat_a_open_sweep_N9_non-reg.png "irs_DirPat_a_open_sweep_N9_non-reg")
+
+`irs_DirPat_a_open_sweep_N9_reg.mat`
+
+![irs_DirPat_a_open_sweep_N9_reg](irs_DirPat_a_open_sweep_N9_reg.png "irs_DirPat_a_open_sweep_N9_reg")
+
+`irs_DirPat_a_wide_sweep_N9_non-reg.mat`
+
+![irs_DirPat_a_wide_sweep_N9_non-reg](irs_DirPat_a_wide_sweep_N9_non-reg.png "irs_DirPat_a_wide_sweep_N9_non-reg")
+
+`irs_DirPat_a_wide_sweep_N9_reg.mat`
+
+![irs_DirPat_a_wide_sweep_N9_reg](irs_DirPat_a_wide_sweep_N9_reg.png "irs_DirPat_a_wide_sweep_N9_reg")
+

README.md

@@ -1,2 +1,18 @@
-# sound-source-directivities
-Data base of spherical harmonic models of sound source directivities
+<a rel="license" href="http://creativecommons.org/licenses/by-nc-sa/4.0/"><img alt="Creative Commons License" style="border-width:0" src="https://i.creativecommons.org/l/by-nc-sa/4.0/88x31.png" /></a><br />This work is licensed under a <a rel="license" href="http://creativecommons.org/licenses/by-nc-sa/4.0/">Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License</a>.
+
+# Database of spherical harmonic models of sound source directivities
+
+This is a database of complete spherical harmonic models of the directivities of sound sources. The data are provided as impulse responses that represent the directivity of the given source in a given discrete direction. The Matlab script `compute_spherical_harmonics_model.m` demonstrates how a spherical harmonic representation can be computed from the data. We do not provide spherical harmonic coefficients directly because of the multitude of definitions of spherical harmonics and also of the Discrete Fourier transform. We rather ask you to select the combination of definitions you would like to use and compute the spherical harmonic coefficients on demand. You may want to add resampling or zero padding and the like to make the data compatible with your intended application. 
+
+As of now, all spherical harmonic models are based on previously published data. Please do not forget to site this repository as well as the original repositories when using the data. References to the original sources are provided with each dataset. All data are limited to the spherical harmonic order `N` that is specified in the corresponding file name. The conversion between raw data and spherical harmonic coefficients is therefore essentially lossless. 
+
+All data are far-field models as described in
+
+> J. Ahrens and S. Bilbao, “Interpolation and Range Extrapolation of Sound Source Directivity Based on a Spherical Wave Propagation Model,” in Proc. of IEEE ICASSP, Barcelona, Spain, May 2020
+
+and
+
+> J. Ahrens and S. Bilbao, “Computation of Spherical Harmonics Based Sound Source Directivity Models from Sparse Measurement Data,” Forum Acusticum, Lyon, France, May 2020.
+
+We will extend and update the data continuously.
+

_tools/asslegendre.m

@@ -0,0 +1,41 @@
+function [Lnm] = asslegendre(n, m, arg)
+%ASSLEGENDRE Calculates the associated Legendre function of degree n and 
+% order m
+%
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+% This work is supplementary material for the book                        %
+%                                                                         %
+% Jens Ahrens, Analytic Methods of Sound Field Synthesis, Springer-Verlag %
+% Berlin Heidelberg, 2012, http://dx.doi.org/10.1007/978-3-642-25743-8    %
+%                                                                         %
+% It has been downloaded from http://soundfieldsynthesis.org and is       %
+% licensed under a Creative Commons Attribution-NonCommercial-ShareAlike  %
+% 3.0 Unported License. Please cite the book appropriately if you use     %
+% these materials in your own work.                                       %
+%                                                                         %
+% (c) 2012 by Jens Ahrens                                                 %
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+
+if (n < 0)
+    error('Degree(n) must not be negative.')
+end
+
+if (n < abs(m))
+    % warning('Absolute value of order(m) must be less than or equal to the degree(n).'); 
+    Lnm = zeros(size(arg));
+    return;
+end
+
+Lnm = legendre(n, arg);
+
+if (n ~= 0)
+    Lnm = squeeze(Lnm(abs(m) + 1, :, :));  
+end
+
+if (m < 0)
+    Lnm = (-1).^abs(m) .* factorial(n - abs(m)) ./ factorial(n + abs(m)) .* Lnm;
+end
+
+Lnm = reshape(Lnm, size(arg));
+
+end

_tools/balloon_plot.m

@@ -0,0 +1,108 @@
+function [] = balloon_plot(coefficients, order, fs, sph_definition)
+% Creates a ballon plot from spherical harmonics coefficients
+
+% plot 4 different frequencies
+f_to_plot    = [500 1000 2000 4000]; % in Hz
+f_to_plot    = f_to_plot(f_to_plot < fs/2); % sort out
+bins_to_plot = round(f_to_plot/(fs/2) * size(coefficients, 1));
+
+% update values
+f_to_plot = bins_to_plot/size(coefficients, 1) * fs/2;
+
+% remove all unused data
+coefficients = coefficients(bins_to_plot, :);
+
+% set up a spatial grid 
+resolution        = 2*order + 2;
+alpha_v           = linspace(0, 2*pi, 2*resolution);  % azimuth
+beta_v            = linspace(0,   pi,   resolution+1);  % colatitude
+[alpha_m, beta_m] = meshgrid(alpha_v, beta_v);
+
+colatitude = beta_m(:).';
+azimuth = alpha_m(:).';
+
+% compute the directivity on the spatial grid
+D = zeros(size(coefficients, 1), size(azimuth, 2));
+
+for l = 0 : order
+    for m = -l : l
+        D = D + coefficients(:, l^2+l+m+1) .* repmat(sphharm(l, m, colatitude, azimuth, sph_definition), [length(bins_to_plot) 1]); 
+    end
+end
+
+% prepare data for plotting
+alpha_m = reshape(azimuth,    resolution+1, []);
+beta_m  = reshape(colatitude, resolution+1, []);
+
+%  --------------------- finally, plot data -------------------------------
+figure;
+set(gcf, 'Color', [1 1 1]);
+
+subplot(2, 2, 1);
+plot_it(abs(reshape(D(1, :), resolution+1, [])),  alpha_m, beta_m, f_to_plot(1));
+
+if length(f_to_plot) > 1
+    subplot(2, 2, 2);
+    plot_it(abs(reshape(D(2, :), resolution+1, [])),  alpha_m, beta_m, f_to_plot(2));
+end
+
+if length(f_to_plot) > 2
+    subplot(2, 2, 3);
+    plot_it(abs(reshape(D(3, :), resolution+1, [])),  alpha_m, beta_m, f_to_plot(3));
+end
+
+if length(f_to_plot) > 3
+    subplot(2, 2, 4);
+    plot_it(abs(reshape(D(4, :), resolution+1, [])),  alpha_m, beta_m, f_to_plot(4));
+end
+
+end
+
+function plot_it(data_to_plot, alpha_m, beta_m, f)
+
+view_angle = [135 20];
+
+color1 = [0.9769    0.9839    0.0805];
+color2 = [0.2440    0.4358    0.9988];
+color3 = [0.2422    0.1504    0.6603];
+
+% avoid a hole in the hull
+alpha_m      = [alpha_m, alpha_m(:, 1)];
+beta_m       = [beta_m, beta_m(:, 1)];
+data_to_plot = [data_to_plot, data_to_plot(:, 1)];
+
+[Xm, Ym, Zm] = sph2cart(alpha_m, pi/2-beta_m, data_to_plot);
+
+surf(Xm, Ym, Zm); 
+
+plot_max = max(abs([Xm(:); Ym(:); Zm(:)]));
+
+% plot coordinate axes
+hold on;
+line([-1 1] * plot_max * .75, [.0 .0], [ 0 0], 'Marker', '.', 'LineStyle', '-', 'Color', [.5 .5 .5], 'LineWidth', 1 );
+line([ 0 0], [-1  1] * plot_max * .75, [ 0 0], 'Marker', '.', 'LineStyle', '-', 'Color', [.5 .5 .5], 'LineWidth', 1 );
+line([ 0 0], [ 0  0], [-1 1] * plot_max * .75, 'Marker', '.', 'LineStyle', '-', 'Color', [.5 .5 .5], 'LineWidth', 1 );
+hold off;
+
+title(sprintf('f = %d Hz', round(f)));
+
+view(view_angle(1), view_angle(2));
+
+box on;
+
+xlabel( '$x$', 'Interpreter', 'latex' );
+ylabel( '$y$', 'Interpreter', 'latex' );
+zlabel( '$z$', 'Interpreter', 'latex' );
+
+axis equal
+axis([-1 1 -1 1 -1 1] * plot_max);
+
+lighting phong
+shading interp
+
+light('Position', [1 0 0], 'Color', color1);
+light('Position', [0 1 0], 'Color', color2);
+light('Position', [0 0 1], 'Color', color3);
+set(gca, 'Projection', 'Perspective', 'FontSize', 7);
+
+end

_tools/dft.m

@@ -0,0 +1,16 @@
+function [tfs] = dft(irs, dft_definition)
+%DFT Discrete Fourier transform
+%   Is always performed along the first dimension of irs
+%   dft_definition: 'matlab' (negative exponent in forward transform) or 
+%                   'williams' (positive exponent in forward transform)
+
+if strcmp(dft_definition, 'matlab')
+    tfs = fft(irs, [], 1);
+elseif strcmp(dft_definition, 'williams')
+    tfs = ifft(irs) * size(irs, 1);
+else 
+    warning('Unknown method.');
+end
+
+end
+

_tools/least_squares_sh_fit.m

@@ -0,0 +1,53 @@
+function [coefficients] = least_squares_sh_fit(N, data, azimuth, colatitude, sphharm_type, reg_parameter)
+% Performs an Nth-order least squares fit of spherical harmonics on data 
+% that is available in the directions (azimuth, colatitude) for each 
+% frequency bin separately using the pseudo inverse
+%
+%   data: frequency goes downwards, position to the right
+%   azimuth, colatitude: row vectors in radians
+%   sphharm_type: see file sphharm.m for the options
+%   reg_parameter (optional): regularization parameters according to 
+%                             Eq. (XX) in XXX; defaults to 0
+%
+%   coefficients: frequency goes downwards, order to the right (sorted
+%                 according to (n^2+n+m+1)
+%
+% Author: Jens Ahrens, March 2020
+
+if nargin < 6
+    reg_parameter = 0;
+end
+
+azimuth    = azimuth.';
+colatitude = colatitude.';
+data       = data.';
+
+% generate a series of n: (0, 1, 1, 1, 2, 2, 2, 2, 2, 3, ...)
+ns = zeros(0, 1);
+
+for n = 0 : N
+    for m = -n : n
+        ns = [ns; n];
+    end
+end
+
+% -------- compute basis functions ----------
+Y_nm = zeros(size(azimuth, 1), (N+1)^2);
+
+for n = 0 : N
+    for m = -n : n
+        Y_nm(:, n^2+n+m+1) = sphharm(n, m, colatitude, azimuth, sphharm_type);
+    end
+end
+
+% ------------- perform the ls fit ---------------
+fprintf('Performing LS fit... ');
+
+coefficients = inv(Y_nm' * Y_nm + reg_parameter * diag(1 + ns .* (ns + 1))) * Y_nm' * data;
+
+fprintf('done.\n');
+
+coefficients = coefficients.';
+
+end
+