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 Computer, 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 "platform/audio/HRTFKernel.h"
34
35 #include "platform/audio/AudioChannel.h"
36 #include "platform/FloatConversion.h"
37 #include "wtf/MathExtras.h"
38
39 namespace blink {
40
41 // Takes the input AudioChannel as an input impulse response and calculates the average group delay.
42 // This represents the initial delay before the most energetic part of the impulse response.
43 // The sample-frame delay is removed from the impulseP impulse response, and this value is returned.
44 // the length of the passed in AudioChannel must be a power of 2.
extractAverageGroupDelay(AudioChannel * channel,size_t analysisFFTSize)45 static float extractAverageGroupDelay(AudioChannel* channel, size_t analysisFFTSize)
46 {
47 ASSERT(channel);
48
49 float* impulseP = channel->mutableData();
50
51 bool isSizeGood = channel->length() >= analysisFFTSize;
52 ASSERT(isSizeGood);
53 if (!isSizeGood)
54 return 0;
55
56 // Check for power-of-2.
57 ASSERT(1UL << static_cast<unsigned>(log2(analysisFFTSize)) == analysisFFTSize);
58
59 FFTFrame estimationFrame(analysisFFTSize);
60 estimationFrame.doFFT(impulseP);
61
62 float frameDelay = narrowPrecisionToFloat(estimationFrame.extractAverageGroupDelay());
63 estimationFrame.doInverseFFT(impulseP);
64
65 return frameDelay;
66 }
67
HRTFKernel(AudioChannel * channel,size_t fftSize,float sampleRate)68 HRTFKernel::HRTFKernel(AudioChannel* channel, size_t fftSize, float sampleRate)
69 : m_frameDelay(0)
70 , m_sampleRate(sampleRate)
71 {
72 ASSERT(channel);
73
74 // Determine the leading delay (average group delay) for the response.
75 m_frameDelay = extractAverageGroupDelay(channel, fftSize / 2);
76
77 float* impulseResponse = channel->mutableData();
78 size_t responseLength = channel->length();
79
80 // We need to truncate to fit into 1/2 the FFT size (with zero padding) in order to do proper convolution.
81 size_t truncatedResponseLength = std::min(responseLength, fftSize / 2); // truncate if necessary to max impulse response length allowed by FFT
82
83 // Quick fade-out (apply window) at truncation point
84 unsigned numberOfFadeOutFrames = static_cast<unsigned>(sampleRate / 4410); // 10 sample-frames @44.1KHz sample-rate
85 ASSERT(numberOfFadeOutFrames < truncatedResponseLength);
86 if (numberOfFadeOutFrames < truncatedResponseLength) {
87 for (unsigned i = truncatedResponseLength - numberOfFadeOutFrames; i < truncatedResponseLength; ++i) {
88 float x = 1.0f - static_cast<float>(i - (truncatedResponseLength - numberOfFadeOutFrames)) / numberOfFadeOutFrames;
89 impulseResponse[i] *= x;
90 }
91 }
92
93 m_fftFrame = adoptPtr(new FFTFrame(fftSize));
94 m_fftFrame->doPaddedFFT(impulseResponse, truncatedResponseLength);
95 }
96
createImpulseResponse()97 PassOwnPtr<AudioChannel> HRTFKernel::createImpulseResponse()
98 {
99 OwnPtr<AudioChannel> channel = adoptPtr(new AudioChannel(fftSize()));
100 FFTFrame fftFrame(*m_fftFrame);
101
102 // Add leading delay back in.
103 fftFrame.addConstantGroupDelay(m_frameDelay);
104 fftFrame.doInverseFFT(channel->mutableData());
105
106 return channel.release();
107 }
108
109 // Interpolates two kernels with x: 0 -> 1 and returns the result.
createInterpolatedKernel(HRTFKernel * kernel1,HRTFKernel * kernel2,float x)110 PassRefPtr<HRTFKernel> HRTFKernel::createInterpolatedKernel(HRTFKernel* kernel1, HRTFKernel* kernel2, float x)
111 {
112 ASSERT(kernel1 && kernel2);
113 if (!kernel1 || !kernel2)
114 return nullptr;
115
116 ASSERT(x >= 0.0 && x < 1.0);
117 x = std::min(1.0f, std::max(0.0f, x));
118
119 float sampleRate1 = kernel1->sampleRate();
120 float sampleRate2 = kernel2->sampleRate();
121 ASSERT(sampleRate1 == sampleRate2);
122 if (sampleRate1 != sampleRate2)
123 return nullptr;
124
125 float frameDelay = (1 - x) * kernel1->frameDelay() + x * kernel2->frameDelay();
126
127 OwnPtr<FFTFrame> interpolatedFrame = FFTFrame::createInterpolatedFrame(*kernel1->fftFrame(), *kernel2->fftFrame(), x);
128 return HRTFKernel::create(interpolatedFrame.release(), frameDelay, sampleRate1);
129 }
130
131 } // namespace blink
132
133 #endif // ENABLE(WEB_AUDIO)
134