Chromium Code Reviews
chromiumcodereview-hr@appspot.gserviceaccount.com (chromiumcodereview-hr) | Please choose your nickname with Settings | Help | Chromium Project | Gerrit Changes | Sign out
(429)

Side by Side Diff: webrtc/modules/audio_processing/intelligibility/intelligibility_enhancer.h

Issue 1207353002: Add new variance update option and unittests for intelligibility (Closed) Base URL: https://chromium.googlesource.com/external/webrtc.git@master
Patch Set: Addressed comments from hlundin Created 5 years, 5 months ago
Use n/p to move between diff chunks; N/P to move between comments. Draft comments are only viewable by you.
Jump to:
View unified diff | Download patch
OLDNEW
1 /* 1 /*
2 * Copyright (c) 2014 The WebRTC project authors. All Rights Reserved. 2 * Copyright (c) 2014 The WebRTC project authors. All Rights Reserved.
hlundin-webrtc 2015/07/02 10:53:13 2015 In fact, I believe all the files in webrtc/m
ekm 2015/07/07 21:57:02 intelligibility_enhancer, _utils, and _proc existe
3 * 3 *
4 * Use of this source code is governed by a BSD-style license 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 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 6 * tree. An additional intellectual property rights grant can be found
7 * in the file PATENTS. All contributing project authors may 7 * in the file PATENTS. All contributing project authors may
8 * be found in the AUTHORS file in the root of the source tree. 8 * be found in the AUTHORS file in the root of the source tree.
9 */ 9 */
10 10
11 // 11 //
12 // Specifies core class for intelligbility enhancement. 12 // Specifies core class for intelligbility enhancement.
(...skipping 63 matching lines...) Expand 10 before | Expand all | Expand 10 after
76 int in_channels, 76 int in_channels,
77 int frames, 77 int frames,
78 int out_channels, 78 int out_channels,
79 std::complex<float>* const* out_block); 79 std::complex<float>* const* out_block);
80 80
81 private: 81 private:
82 IntelligibilityEnhancer* parent_; 82 IntelligibilityEnhancer* parent_;
83 AudioSource source_; 83 AudioSource source_;
84 }; 84 };
85 friend class TransformCallback; 85 friend class TransformCallback;
86 FRIEND_TEST_ALL_PREFIXES(IntelligibilityEnhancerTest, TestErbCreation);
87 FRIEND_TEST_ALL_PREFIXES(IntelligibilityEnhancerTest, TestSolveForGains);
86 88
87 // Sends streams to ProcessClearBlock or ProcessNoiseBlock based on source. 89 // Sends streams to ProcessClearBlock or ProcessNoiseBlock based on source.
88 void DispatchAudio(AudioSource source, 90 void DispatchAudio(AudioSource source,
89 const std::complex<float>* in_block, 91 const std::complex<float>* in_block,
90 std::complex<float>* out_block); 92 std::complex<float>* out_block);
91 93
92 // Updates variance computation and analysis with |in_block_|, 94 // Updates variance computation and analysis with |in_block_|,
93 // and writes modified speech to |out_block|. 95 // and writes modified speech to |out_block|.
94 void ProcessClearBlock(const std::complex<float>* in_block, 96 void ProcessClearBlock(const std::complex<float>* in_block,
95 std::complex<float>* out_block); 97 std::complex<float>* out_block);
96 98
97 // Computes and sets modified gains. 99 // Computes and sets modified gains.
98 void AnalyzeClearBlock(float power_target); 100 void AnalyzeClearBlock(float power_target);
99 101
102 // Bisection search for optimal |lambda|.
103 void SolveForLambda(float power_target, float power_bot, float power_top);
104
105 // Transforms freq gains to ERB gains.
106 void UpdateErbGains();
107
100 // Updates variance calculation for noise input with |in_block|. 108 // Updates variance calculation for noise input with |in_block|.
101 void ProcessNoiseBlock(const std::complex<float>* in_block, 109 void ProcessNoiseBlock(const std::complex<float>* in_block,
102 std::complex<float>* out_block); 110 std::complex<float>* out_block);
103 111
104 // Returns number of ERB filters. 112 // Returns number of ERB filters.
105 static int GetBankSize(int sample_rate, int erb_resolution); 113 static int GetBankSize(int sample_rate, int erb_resolution);
106 114
107 // Initializes ERB filterbank. 115 // Initializes ERB filterbank.
108 void CreateErbBank(); 116 void CreateErbBank();
109 117
110 // Analytically solves quadratic for optimal gains given |lambda|. 118 // Analytically solves quadratic for optimal gains given |lambda|.
111 // Negative gains are set to 0. Stores the results in |sols|. 119 // Negative gains are set to 0. Stores the results in |sols|.
112 void SolveForGainsGivenLambda(float lambda, int start_freq, float* sols); 120 void SolveForGainsGivenLambda(float lambda, int start_freq, float* sols);
113 121
114 // Computes variance across ERB filters from freq variance |var|. 122 // Computes variance across ERB filters from freq variance |var|.
115 // Stores in |result|. 123 // Stores in |result|.
116 void FilterVariance(const float* var, float* result); 124 void FilterVariance(const float* var, float* result);
117 125
118 // Returns dot product of vectors specified by size |length| arrays |a|,|b|. 126 // Returns dot product of vectors specified by size |length| arrays |a|,|b|.
119 static float DotProduct(const float* a, const float* b, int length); 127 static float DotProduct(const float* a, const float* b, int length);
120 128
121 static const int kErbResolution;
122 static const int kWindowSizeMs;
123 static const int kChunkSizeMs;
124 static const int kAnalyzeRate; // Default for |analysis_rate_|.
125 static const int kVarianceRate; // Default for |variance_rate_|.
126 static const float kClipFreq;
127 static const float kConfigRho; // Default production and interpretation SNR.
128 static const float kKbdAlpha;
129 static const float kGainChangeLimit;
130
131 const int freqs_; // Num frequencies in frequency domain. 129 const int freqs_; // Num frequencies in frequency domain.
132 const int window_size_; // Window size in samples; also the block size. 130 const int window_size_; // Window size in samples; also the block size.
133 const int chunk_length_; // Chunk size in samples. 131 const int chunk_length_; // Chunk size in samples.
134 const int bank_size_; // Num ERB filters. 132 const int bank_size_; // Num ERB filters.
135 const int sample_rate_hz_; 133 const int sample_rate_hz_;
136 const int erb_resolution_; 134 const int erb_resolution_;
137 const int channels_; // Num channels. 135 const int channels_; // Num channels.
138 const int analysis_rate_; // Num blocks before gains recalculated. 136 const int analysis_rate_; // Num blocks before gains recalculated.
139 const int variance_rate_; // Num recalculations before history is cleared. 137 const int variance_rate_; // Num recalculations before history is cleared.
140 138
(...skipping 28 matching lines...) Expand all
169 // Note: VAD currently does not affect anything in IntelligibilityEnhancer. 167 // Note: VAD currently does not affect anything in IntelligibilityEnhancer.
170 VadInst* vad_high_; 168 VadInst* vad_high_;
171 VadInst* vad_low_; 169 VadInst* vad_low_;
172 rtc::scoped_ptr<int16_t[]> vad_tmp_buffer_; 170 rtc::scoped_ptr<int16_t[]> vad_tmp_buffer_;
173 bool has_voice_low_; // Whether voice detected in speech stream. 171 bool has_voice_low_; // Whether voice detected in speech stream.
174 }; 172 };
175 173
176 } // namespace webrtc 174 } // namespace webrtc
177 175
178 #endif // WEBRTC_MODULES_AUDIO_PROCESSING_INTELLIGIBILITY_INTELLIGIBILITY_ENHAN CER_H_ 176 #endif // WEBRTC_MODULES_AUDIO_PROCESSING_INTELLIGIBILITY_INTELLIGIBILITY_ENHAN CER_H_
OLDNEW

Powered by Google App Engine
This is Rietveld 408576698