OLD | NEW |
1 /* | 1 /* |
2 * Copyright (c) 2014 The WebRTC project authors. All Rights Reserved. | 2 * Copyright (c) 2014 The WebRTC project authors. All Rights Reserved. |
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 #ifndef WEBRTC_MODULES_AUDIO_PROCESSING_INTELLIGIBILITY_INTELLIGIBILITY_ENHANCER
_H_ | 11 #ifndef WEBRTC_MODULES_AUDIO_PROCESSING_INTELLIGIBILITY_INTELLIGIBILITY_ENHANCER
_H_ |
12 #define WEBRTC_MODULES_AUDIO_PROCESSING_INTELLIGIBILITY_INTELLIGIBILITY_ENHANCER
_H_ | 12 #define WEBRTC_MODULES_AUDIO_PROCESSING_INTELLIGIBILITY_INTELLIGIBILITY_ENHANCER
_H_ |
13 | 13 |
14 #include <complex> | 14 #include <complex> |
15 #include <memory> | 15 #include <memory> |
16 #include <vector> | 16 #include <vector> |
17 | 17 |
18 #include "webrtc/common_audio/lapped_transform.h" | 18 #include "webrtc/common_audio/lapped_transform.h" |
19 #include "webrtc/common_audio/channel_buffer.h" | 19 #include "webrtc/common_audio/channel_buffer.h" |
| 20 #include "webrtc/common_audio/swap_queue.h" |
20 #include "webrtc/modules/audio_processing/intelligibility/intelligibility_utils.
h" | 21 #include "webrtc/modules/audio_processing/intelligibility/intelligibility_utils.
h" |
| 22 #include "webrtc/modules/audio_processing/processing_component.h" |
21 #include "webrtc/modules/audio_processing/vad/voice_activity_detector.h" | 23 #include "webrtc/modules/audio_processing/vad/voice_activity_detector.h" |
22 | 24 |
23 namespace webrtc { | 25 namespace webrtc { |
24 | 26 |
25 // Speech intelligibility enhancement module. Reads render and capture | 27 // Speech intelligibility enhancement module. Reads render and capture |
26 // audio streams and modifies the render stream with a set of gains per | 28 // audio streams and modifies the render stream with a set of gains per |
27 // frequency bin to enhance speech against the noise background. | 29 // frequency bin to enhance speech against the noise background. |
28 // Details of the model and algorithm can be found in the original paper: | 30 // Details of the model and algorithm can be found in the original paper: |
29 // http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=6882788 | 31 // http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=6882788 |
30 class IntelligibilityEnhancer : public LappedTransform::Callback { | 32 class IntelligibilityEnhancer : public LappedTransform::Callback { |
31 public: | 33 public: |
32 IntelligibilityEnhancer(int sample_rate_hz, size_t num_render_channels); | 34 IntelligibilityEnhancer(int sample_rate_hz, |
| 35 size_t num_render_channels, |
| 36 size_t num_noise_bins); |
33 | 37 |
34 // Sets the capture noise magnitude spectrum estimate. | 38 // Sets the capture noise magnitude spectrum estimate. |
35 void SetCaptureNoiseEstimate(std::vector<float> noise); | 39 void SetCaptureNoiseEstimate(std::vector<float> noise); |
36 | 40 |
37 // Reads chunk of speech in time domain and updates with modified signal. | 41 // Reads chunk of speech in time domain and updates with modified signal. |
38 void ProcessRenderAudio(float* const* audio, | 42 void ProcessRenderAudio(float* const* audio, |
39 int sample_rate_hz, | 43 int sample_rate_hz, |
40 size_t num_channels); | 44 size_t num_channels); |
41 bool active() const; | 45 bool active() const; |
42 | 46 |
(...skipping 22 matching lines...) Expand all Loading... |
65 // Initializes ERB filterbank. | 69 // Initializes ERB filterbank. |
66 std::vector<std::vector<float>> CreateErbBank(size_t num_freqs); | 70 std::vector<std::vector<float>> CreateErbBank(size_t num_freqs); |
67 | 71 |
68 // Analytically solves quadratic for optimal gains given |lambda|. | 72 // Analytically solves quadratic for optimal gains given |lambda|. |
69 // Negative gains are set to 0. Stores the results in |sols|. | 73 // Negative gains are set to 0. Stores the results in |sols|. |
70 void SolveForGainsGivenLambda(float lambda, size_t start_freq, float* sols); | 74 void SolveForGainsGivenLambda(float lambda, size_t start_freq, float* sols); |
71 | 75 |
72 // Returns true if the audio is speech. | 76 // Returns true if the audio is speech. |
73 bool IsSpeech(const float* audio); | 77 bool IsSpeech(const float* audio); |
74 | 78 |
| 79 static const size_t kMaxNumNoiseEstimatesToBuffer = 5; |
| 80 |
75 const size_t freqs_; // Num frequencies in frequency domain. | 81 const size_t freqs_; // Num frequencies in frequency domain. |
| 82 const size_t num_noise_bins_; |
76 const size_t chunk_length_; // Chunk size in samples. | 83 const size_t chunk_length_; // Chunk size in samples. |
77 const size_t bank_size_; // Num ERB filters. | 84 const size_t bank_size_; // Num ERB filters. |
78 const int sample_rate_hz_; | 85 const int sample_rate_hz_; |
79 const size_t num_render_channels_; | 86 const size_t num_render_channels_; |
80 | 87 |
81 intelligibility::PowerEstimator<std::complex<float>> clear_power_estimator_; | 88 intelligibility::PowerEstimator<std::complex<float>> clear_power_estimator_; |
82 std::unique_ptr<intelligibility::PowerEstimator<float>> | 89 intelligibility::PowerEstimator<float> noise_power_estimator_; |
83 noise_power_estimator_; | |
84 std::vector<float> filtered_clear_pow_; | 90 std::vector<float> filtered_clear_pow_; |
85 std::vector<float> filtered_noise_pow_; | 91 std::vector<float> filtered_noise_pow_; |
86 std::vector<float> center_freqs_; | 92 std::vector<float> center_freqs_; |
87 std::vector<std::vector<float>> capture_filter_bank_; | 93 std::vector<std::vector<float>> capture_filter_bank_; |
88 std::vector<std::vector<float>> render_filter_bank_; | 94 std::vector<std::vector<float>> render_filter_bank_; |
89 size_t start_freq_; | 95 size_t start_freq_; |
90 | 96 |
91 std::vector<float> gains_eq_; // Pre-filter modified gains. | 97 std::vector<float> gains_eq_; // Pre-filter modified gains. |
92 intelligibility::GainApplier gain_applier_; | 98 intelligibility::GainApplier gain_applier_; |
93 | 99 |
94 std::unique_ptr<LappedTransform> render_mangler_; | 100 std::unique_ptr<LappedTransform> render_mangler_; |
95 | 101 |
96 VoiceActivityDetector vad_; | 102 VoiceActivityDetector vad_; |
97 std::vector<int16_t> audio_s16_; | 103 std::vector<int16_t> audio_s16_; |
98 size_t chunks_since_voice_; | 104 size_t chunks_since_voice_; |
99 bool is_speech_; | 105 bool is_speech_; |
| 106 |
| 107 std::vector<float> noise_estimation_buffer_; |
| 108 SwapQueue<std::vector<float>, RenderQueueItemVerifier<float>> |
| 109 noise_estimation_queue_; |
100 }; | 110 }; |
101 | 111 |
102 } // namespace webrtc | 112 } // namespace webrtc |
103 | 113 |
104 #endif // WEBRTC_MODULES_AUDIO_PROCESSING_INTELLIGIBILITY_INTELLIGIBILITY_ENHAN
CER_H_ | 114 #endif // WEBRTC_MODULES_AUDIO_PROCESSING_INTELLIGIBILITY_INTELLIGIBILITY_ENHAN
CER_H_ |
OLD | NEW |