1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
|
/*
* Copyright (c) 2018 The WebRTC project authors. All Rights Reserved.
*
* Use of this source code is governed by a BSD-style license
* that can be found in the LICENSE file in the root of the source
* tree. An additional intellectual property rights grant can be found
* in the file PATENTS. All contributing project authors may
* be found in the AUTHORS file in the root of the source tree.
*/
#include "modules/audio_processing/agc2/adaptive_agc.h"
#include "common_audio/include/audio_util.h"
#include "modules/audio_processing/agc2/cpu_features.h"
#include "modules/audio_processing/agc2/vad_with_level.h"
#include "modules/audio_processing/logging/apm_data_dumper.h"
#include "rtc_base/checks.h"
#include "rtc_base/logging.h"
namespace webrtc {
namespace {
using AdaptiveDigitalConfig =
AudioProcessing::Config::GainController2::AdaptiveDigital;
using NoiseEstimatorType =
AudioProcessing::Config::GainController2::NoiseEstimator;
constexpr int kGainApplierAdjacentSpeechFramesThreshold = 1;
constexpr float kMaxGainChangePerSecondDb = 3.0f;
constexpr float kMaxOutputNoiseLevelDbfs = -50.0f;
// Detects the available CPU features and applies any kill-switches.
AvailableCpuFeatures GetAllowedCpuFeatures(
const AdaptiveDigitalConfig& config) {
AvailableCpuFeatures features = GetAvailableCpuFeatures();
if (!config.sse2_allowed) {
features.sse2 = false;
}
if (!config.avx2_allowed) {
features.avx2 = false;
}
if (!config.neon_allowed) {
features.neon = false;
}
return features;
}
std::unique_ptr<NoiseLevelEstimator> CreateNoiseLevelEstimator(
NoiseEstimatorType estimator_type,
ApmDataDumper* apm_data_dumper) {
switch (estimator_type) {
case NoiseEstimatorType::kStationaryNoise:
return CreateStationaryNoiseEstimator(apm_data_dumper);
case NoiseEstimatorType::kNoiseFloor:
return CreateNoiseFloorEstimator(apm_data_dumper);
}
}
constexpr NoiseEstimatorType kDefaultNoiseLevelEstimatorType =
NoiseEstimatorType::kNoiseFloor;
} // namespace
AdaptiveAgc::AdaptiveAgc(ApmDataDumper* apm_data_dumper)
: speech_level_estimator_(apm_data_dumper),
gain_controller_(apm_data_dumper,
kGainApplierAdjacentSpeechFramesThreshold,
kMaxGainChangePerSecondDb,
kMaxOutputNoiseLevelDbfs),
apm_data_dumper_(apm_data_dumper),
noise_level_estimator_(
CreateNoiseLevelEstimator(kDefaultNoiseLevelEstimatorType,
apm_data_dumper)),
saturation_protector_(
CreateSaturationProtector(kSaturationProtectorInitialHeadroomDb,
kSaturationProtectorExtraHeadroomDb,
kGainApplierAdjacentSpeechFramesThreshold,
apm_data_dumper)) {
RTC_DCHECK(apm_data_dumper);
}
AdaptiveAgc::AdaptiveAgc(ApmDataDumper* apm_data_dumper,
const AdaptiveDigitalConfig& config)
: speech_level_estimator_(apm_data_dumper,
config.adjacent_speech_frames_threshold),
vad_(config.vad_reset_period_ms, GetAllowedCpuFeatures(config)),
gain_controller_(apm_data_dumper,
config.adjacent_speech_frames_threshold,
config.max_gain_change_db_per_second,
config.max_output_noise_level_dbfs),
apm_data_dumper_(apm_data_dumper),
noise_level_estimator_(
CreateNoiseLevelEstimator(config.noise_estimator, apm_data_dumper)),
saturation_protector_(
CreateSaturationProtector(kSaturationProtectorInitialHeadroomDb,
kSaturationProtectorExtraHeadroomDb,
config.adjacent_speech_frames_threshold,
apm_data_dumper)) {
RTC_DCHECK(apm_data_dumper);
RTC_DCHECK(noise_level_estimator_);
RTC_DCHECK(saturation_protector_);
if (!config.use_saturation_protector) {
RTC_LOG(LS_WARNING) << "The saturation protector cannot be disabled.";
}
}
AdaptiveAgc::~AdaptiveAgc() = default;
void AdaptiveAgc::Process(AudioFrameView<float> frame, float limiter_envelope) {
AdaptiveDigitalGainApplier::FrameInfo info;
VadLevelAnalyzer::Result vad_result = vad_.AnalyzeFrame(frame);
info.speech_probability = vad_result.speech_probability;
apm_data_dumper_->DumpRaw("agc2_speech_probability",
vad_result.speech_probability);
apm_data_dumper_->DumpRaw("agc2_input_rms_dbfs", vad_result.rms_dbfs);
apm_data_dumper_->DumpRaw("agc2_input_peak_dbfs", vad_result.peak_dbfs);
speech_level_estimator_.Update(vad_result);
info.speech_level_dbfs = speech_level_estimator_.level_dbfs();
info.speech_level_reliable = speech_level_estimator_.IsConfident();
apm_data_dumper_->DumpRaw("agc2_speech_level_dbfs", info.speech_level_dbfs);
apm_data_dumper_->DumpRaw("agc2_speech_level_reliable",
info.speech_level_reliable);
info.noise_rms_dbfs = noise_level_estimator_->Analyze(frame);
apm_data_dumper_->DumpRaw("agc2_noise_rms_dbfs", info.noise_rms_dbfs);
saturation_protector_->Analyze(info.speech_probability, vad_result.peak_dbfs,
info.speech_level_dbfs);
info.headroom_db = saturation_protector_->HeadroomDb();
apm_data_dumper_->DumpRaw("agc2_headroom_db", info.headroom_db);
info.limiter_envelope_dbfs = FloatS16ToDbfs(limiter_envelope);
apm_data_dumper_->DumpRaw("agc2_limiter_envelope_dbfs",
info.limiter_envelope_dbfs);
gain_controller_.Process(info, frame);
}
void AdaptiveAgc::HandleInputGainChange() {
speech_level_estimator_.Reset();
saturation_protector_->Reset();
}
} // namespace webrtc
|
