1 /*
2 * Copyright (c) 2019 The WebRTC project authors. All Rights Reserved.
3 *
4 * Use of this source code is governed by a BSD-style license
5 * that can be found in the LICENSE file in the root of the source
6 * tree. An additional intellectual property rights grant can be found
7 * in the file PATENTS. All contributing project authors may
8 * be found in the AUTHORS file in the root of the source tree.
9 */
10
11 #include "modules/audio_processing/voice_detection.h"
12
13 #include "common_audio/vad/include/webrtc_vad.h"
14 #include "modules/audio_processing/audio_buffer.h"
15 #include "rtc_base/checks.h"
16
17 namespace webrtc {
18 class VoiceDetection::Vad {
19 public:
Vad()20 Vad() {
21 state_ = WebRtcVad_Create();
22 RTC_CHECK(state_);
23 int error = WebRtcVad_Init(state_);
24 RTC_DCHECK_EQ(0, error);
25 }
~Vad()26 ~Vad() { WebRtcVad_Free(state_); }
27
28 Vad(Vad&) = delete;
29 Vad& operator=(Vad&) = delete;
30
state()31 VadInst* state() { return state_; }
32
33 private:
34 VadInst* state_ = nullptr;
35 };
36
VoiceDetection(int sample_rate_hz,Likelihood likelihood)37 VoiceDetection::VoiceDetection(int sample_rate_hz, Likelihood likelihood)
38 : sample_rate_hz_(sample_rate_hz),
39 frame_size_samples_(static_cast<size_t>(sample_rate_hz_ / 100)),
40 likelihood_(likelihood),
41 vad_(new Vad()) {
42 int mode = 2;
43 switch (likelihood) {
44 case VoiceDetection::kVeryLowLikelihood:
45 mode = 3;
46 break;
47 case VoiceDetection::kLowLikelihood:
48 mode = 2;
49 break;
50 case VoiceDetection::kModerateLikelihood:
51 mode = 1;
52 break;
53 case VoiceDetection::kHighLikelihood:
54 mode = 0;
55 break;
56 default:
57 RTC_NOTREACHED();
58 break;
59 }
60 int error = WebRtcVad_set_mode(vad_->state(), mode);
61 RTC_DCHECK_EQ(0, error);
62 }
63
~VoiceDetection()64 VoiceDetection::~VoiceDetection() {}
65
ProcessCaptureAudio(AudioBuffer * audio)66 bool VoiceDetection::ProcessCaptureAudio(AudioBuffer* audio) {
67 RTC_DCHECK_GE(AudioBuffer::kMaxSplitFrameLength,
68 audio->num_frames_per_band());
69 std::array<int16_t, AudioBuffer::kMaxSplitFrameLength> mixed_low_pass_data;
70 rtc::ArrayView<const int16_t> mixed_low_pass(mixed_low_pass_data.data(),
71 audio->num_frames_per_band());
72 if (audio->num_channels() == 1) {
73 FloatS16ToS16(audio->split_bands_const(0)[kBand0To8kHz],
74 audio->num_frames_per_band(), mixed_low_pass_data.data());
75 } else {
76 const int num_channels = static_cast<int>(audio->num_channels());
77 for (size_t i = 0; i < audio->num_frames_per_band(); ++i) {
78 int32_t value =
79 FloatS16ToS16(audio->split_channels_const(kBand0To8kHz)[0][i]);
80 for (int j = 1; j < num_channels; ++j) {
81 value += FloatS16ToS16(audio->split_channels_const(kBand0To8kHz)[j][i]);
82 }
83 mixed_low_pass_data[i] = value / num_channels;
84 }
85 }
86
87 int vad_ret = WebRtcVad_Process(vad_->state(), sample_rate_hz_,
88 mixed_low_pass.data(), frame_size_samples_);
89 RTC_DCHECK(vad_ret == 0 || vad_ret == 1);
90 return vad_ret == 0 ? false : true;
91 }
92 } // namespace webrtc
93