diff options
Diffstat (limited to 'util/adt_generate_qt.m')
| -rw-r--r-- | util/adt_generate_qt.m | 165 |
1 files changed, 165 insertions, 0 deletions
diff --git a/util/adt_generate_qt.m b/util/adt_generate_qt.m new file mode 100644 index 000000000..453732273 --- /dev/null +++ b/util/adt_generate_qt.m @@ -0,0 +1,165 @@ +%{ +** Copyright (C) 2021 The Qt Company Ltd. +** SPDX-License-Identifier: LicenseRef-Qt-Commercial OR GPL-3.0-only WITH Qt-GPL-exception-1.0 +%} + +% Create mapping matrixes to convert ambisonic to different speaker layouts supported in Qt Multimedia +% +% This uses the ADT octave framework from https://bitbucket.org/ambidecodertoolbox/adt/src/master/ +% to generate conversion matrixes between ambisonic formats and various speaker configurations +% +% We're generating two band conversion matrices with a cutoff frequency of 380 Herz. Filtering low +% and high frequencies differently is important to get a decent spatial reproduction. For details +% see the "Is my decoder Ambisonic?" paper (https://ambisonics.dreamhosters.com/BLaH3.pdf) +% +function adt_generate_qt() + % assume speakers are 2 meters from listener. This should avoid near-field + % effects and should work for most room setups + radius = 2; + + [outfile,msg] = fopen("qambisonicdecoderdata_p.h",'w'); + fprintf(outfile, "// Copyright (C) 2016 The Qt Company Ltd.\n"); + fprintf(outfile, "//SPDX-License-Identifier: LicenseRef-Qt-Commercial OR LGPL-3.0-only OR GPL-2.0-only OR GPL-3.0-only\n"); + fprintf(outfile, "#ifndef QAMBISONICDECODERDATA_P_H\n"); + fprintf(outfile, "#define QAMBISONICDECODERDATA_P_H\n\n"); + fprintf(outfile, "#include <qtspatialaudioglobal_p.h>\n\n"); + fprintf(outfile, "// W A R N I N G\n"); + fprintf(outfile, "// -------------\n"); + fprintf(outfile, "//\n"); + fprintf(outfile, "// This file is not part of the Qt API. It exists purely as an\n"); + fprintf(outfile, "// implementation detail. This header file may change from version to\n"); + fprintf(outfile, "// version without notice, or even be removed.\n"); + fprintf(outfile, "//\n"); + fprintf(outfile, "// We mean it.\n"); + fprintf(outfile, "//\n\n"); + fprintf(outfile, "// This file is generated by the matlab/octave file adt_generate_qt.m\n"); + fprintf(outfile, "// using the Ambisonic Decoder Toolbox (https://bitbucket.org/ambidecodertoolbox/adt/src/master/)\n"); + fprintf(outfile, "\n\n"); + fprintf(outfile, "QT_BEGIN_NAMESPACE\n\n"); + + % cover top/bottom for surround + imag_speakers = [0,0,radius; 0,0,-radius]; + + % https://www.dolby.com/us/en/guide/surround-sound-speaker-setup/5-1-setup.html + % Dolby 5.1 -- F: 22-30, S: 110-120, C in same plane as F + S = ambi_spkr_array(... + '5dot0', ... + 'AER', 'DDM', ... + 'L', [ 30, 0, radius], ... + 'R', [ -30, 0, radius], ... + 'C', [ 0, 0, radius], ... + 'Ls', [ 110, 0, radius], ... + 'Rs', [-110, 0, radius] ... + ); + createDecoders(S, imag_speakers, outfile); + + S.lfeRow = 4; + S.name = "5dot1"; + createDecoders(S, imag_speakers, outfile); + + % https://www.dolby.com/us/en/guide/surround-sound-speaker-setup/7-1-setup.html + % Dolby 7.1 F: 22-30, S: 90-110, B: 135-150 + S = ambi_spkr_array(... + ... % array name + '7dot0', ... + ... % coordinate codes, unit codes + ... % Azimuth, Elevation, Radius; Degrees, Degrees, Meters + 'AER', 'DDM', ... + ... % speaker name, [azimuth, elevation, radius] + 'L', [ 30, 0, radius], ... + 'R', [ -30, 0, radius], ... + 'C', [ 0, 0, radius], ... + 'Ls', [ 90, 0, radius], ... + 'Rs', [ -90, 0, radius], ... + 'Lb', [ 150, 0, radius], ... + 'Rb', [-150, 0, radius] ... + ); + createDecoders(S, imag_speakers, outfile); + + S.lfeRow = 4; + S.name = "7dot1"; + createDecoders(S, imag_speakers, outfile); + + fprintf(outfile, "QT_END_NAMESPACE\n\n"); + fprintf(outfile, "#endif\n\n"); + fclose(outfile); +end + +% remove rounding errors +function m = trimMatrix(m) + for i = 1:rows(m) + for j = 1:columns(m) + if (abs(m(i, j)) < 1e-4) + m(i, j) = 0; + endif + endfor + endfor +end + +function writeLFERow(outfile, m, suffix) + if (strcmp(suffix, 'hf')) + fprintf(outfile, "0.0f, "); % no need for High frequency data on the LFE channel + else + fprintf(outfile, "0.5f, "); + endif + for i = 2:columns(m) + fprintf(outfile, "0.0f, "); + endfor + fprintf(outfile, "// LFE\n"); +end + +function [n, m] = getnm(l) +% Computes spherical harmonic degree and order from Ambisonic Channel Number. + n = floor(sqrt(l)); + m = l-n.^2-n; +end + +function channels = normalizeSN3D(channels) + for i = 1:columns(channels) + [n, m] = getnm(i-1); + if (m == 0) + factor = 1; + else + factor = sqrt(2 * factorial(n - abs(m)) / (factorial(n + abs(m)))); + endif + % There's an adjustment factor of 1.5 here, to match the volumes generated by the matrices + % here with what we get from a simple stereo decoding. The factor is heuristically determined + channels(i) *= factor*1.5; + endfor +end + +function writeMatrix(outfile, level, S, M, suffix) + m = trimMatrix(M); + hasLFE = isfield(S, "lfeRow"); + r = rows(m); + c = columns(m); + if (hasLFE) + r = r + 1; + endif + fprintf(outfile, "// Decoder matrix for %s, ambisonic level %d\n", S.name, level); + fprintf(outfile, "static constexpr float decoderMatrix_%s_%d_%s[%d*%d] = {\n", S.name, level, suffix, r, c); + for i = 1:rows(S.id) + channels = normalizeSN3D(m(i, :)) + fprintf(outfile, "%ff, ", channels); + fprintf(outfile, "// %s\n", S.id(i, 1){1}); + if (hasLFE && S.lfeRow == i + 1) + writeLFERow(outfile, m, suffix); + endif + endfor + fprintf(outfile, "};\n\n"); +end + +function createOneDecoder(S, imag_speakers, outfile, level) + ambi_order = ambi_channel_definitions_convention(level, 'ambix2011') + [D,S,M,C] = ambi_run_allrad(S, ambi_order, imag_speakers, [S.name '_' int2str(level)], false, "amb", 1, 3); + writeMatrix(outfile, level, S, M.lf, "lf"); + m = ambi_apply_gamma(M.hf, D.hf_gains, C); + writeMatrix(outfile, level, S, m, "hf"); +end + +function createDecoders(S, imag_speakers, outfile) + for level = [1:3] + createOneDecoder(S, imag_speakers, outfile, level) + endfor +end + |