1/*
2 * Copyright (C) 2010 Google Inc. All rights reserved.
3 *
4 * Redistribution and use in source and binary forms, with or without
5 * modification, are permitted provided that the following conditions
6 * are met:
7 *
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
13 * 3. Neither the name of Apple Inc. ("Apple") nor the names of
14 * its contributors may be used to endorse or promote products derived
15 * from this software without specific prior written permission.
16 *
17 * THIS SOFTWARE IS PROVIDED BY APPLE AND ITS CONTRIBUTORS "AS IS" AND ANY
18 * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
19 * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
20 * DISCLAIMED. IN NO EVENT SHALL APPLE OR ITS CONTRIBUTORS BE LIABLE FOR ANY
21 * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
22 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
23 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
24 * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
25 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
26 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
27 */
28
29#include "config.h"
30
31#if ENABLE(WEB_AUDIO)
32
33#include "HRTFKernel.h"
34
35#include "AudioChannel.h"
36#include "Biquad.h"
37#include "FFTFrame.h"
38#include "FloatConversion.h"
39#include <wtf/MathExtras.h>
40
41namespace WebCore {
42
43// Takes the input AudioChannel as an input impulse response and calculates the average group delay.
44// This represents the initial delay before the most energetic part of the impulse response.
45// The sample-frame delay is removed from the impulseP impulse response, and this value is returned.
46// the length of the passed in AudioChannel must be a power of 2.
47static float extractAverageGroupDelay(AudioChannel* channel, size_t analysisFFTSize)
48{
49 ASSERT(channel);
50
51 float* impulseP = channel->mutableData();
52
53 bool isSizeGood = channel->length() >= analysisFFTSize;
54 ASSERT(isSizeGood);
55 if (!isSizeGood)
56 return 0;
57
58 // Check for power-of-2.
59 ASSERT(1UL << static_cast<unsigned>(log2(analysisFFTSize)) == analysisFFTSize);
60
61 FFTFrame estimationFrame(analysisFFTSize);
62 estimationFrame.doFFT(impulseP);
63
64 float frameDelay = narrowPrecisionToFloat(estimationFrame.extractAverageGroupDelay());
65 estimationFrame.doInverseFFT(impulseP);
66
67 return frameDelay;
68}
69
70HRTFKernel::HRTFKernel(AudioChannel* channel, size_t fftSize, float sampleRate)
71 : m_frameDelay(0)
72 , m_sampleRate(sampleRate)
73{
74 ASSERT(channel);
75
76 // Determine the leading delay (average group delay) for the response.
77 m_frameDelay = extractAverageGroupDelay(channel, fftSize / 2);
78
79 float* impulseResponse = channel->mutableData();
80 size_t responseLength = channel->length();
81
82 // We need to truncate to fit into 1/2 the FFT size (with zero padding) in order to do proper convolution.
83 size_t truncatedResponseLength = std::min(responseLength, fftSize / 2); // truncate if necessary to max impulse response length allowed by FFT
84
85 // Quick fade-out (apply window) at truncation point
86 unsigned numberOfFadeOutFrames = static_cast<unsigned>(sampleRate / 4410); // 10 sample-frames @44.1KHz sample-rate
87 ASSERT(numberOfFadeOutFrames < truncatedResponseLength);
88 if (numberOfFadeOutFrames < truncatedResponseLength) {
89 for (unsigned i = truncatedResponseLength - numberOfFadeOutFrames; i < truncatedResponseLength; ++i) {
90 float x = 1.0f - static_cast<float>(i - (truncatedResponseLength - numberOfFadeOutFrames)) / numberOfFadeOutFrames;
91 impulseResponse[i] *= x;
92 }
93 }
94
95 m_fftFrame = std::make_unique<FFTFrame>(fftSize);
96 m_fftFrame->doPaddedFFT(impulseResponse, truncatedResponseLength);
97}
98
99size_t HRTFKernel::fftSize() const
100{
101 return m_fftFrame->fftSize();
102}
103
104std::unique_ptr<AudioChannel> HRTFKernel::createImpulseResponse()
105{
106 auto channel = std::make_unique<AudioChannel>(fftSize());
107 FFTFrame fftFrame(*m_fftFrame);
108
109 // Add leading delay back in.
110 fftFrame.addConstantGroupDelay(m_frameDelay);
111 fftFrame.doInverseFFT(channel->mutableData());
112
113 return channel;
114}
115
116// Interpolates two kernels with x: 0 -> 1 and returns the result.
117RefPtr<HRTFKernel> HRTFKernel::createInterpolatedKernel(HRTFKernel* kernel1, HRTFKernel* kernel2, float x)
118{
119 ASSERT(kernel1 && kernel2);
120 if (!kernel1 || !kernel2)
121 return nullptr;
122
123 ASSERT(x >= 0.0 && x < 1.0);
124 x = std::min(1.0f, std::max(0.0f, x));
125
126 float sampleRate1 = kernel1->sampleRate();
127 float sampleRate2 = kernel2->sampleRate();
128 ASSERT(sampleRate1 == sampleRate2);
129 if (sampleRate1 != sampleRate2)
130 return nullptr;
131
132 float frameDelay = (1 - x) * kernel1->frameDelay() + x * kernel2->frameDelay();
133
134 std::unique_ptr<FFTFrame> interpolatedFrame = FFTFrame::createInterpolatedFrame(*kernel1->fftFrame(), *kernel2->fftFrame(), x);
135 return HRTFKernel::create(WTFMove(interpolatedFrame), frameDelay, sampleRate1);
136}
137
138} // namespace WebCore
139
140#endif // ENABLE(WEB_AUDIO)
141