| OLD | NEW |
| 1 /* | 1 /* |
| 2 * Copyright (c) 2012 The WebRTC project authors. All Rights Reserved. | 2 * Copyright (c) 2012 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 #include "webrtc/modules/audio_coding/neteq/time_stretch.h" | 11 #include "webrtc/modules/audio_coding/neteq/time_stretch.h" |
| 12 | 12 |
| 13 #include <algorithm> // min, max | 13 #include <algorithm> // min, max |
| 14 #include <memory> | 14 #include <memory> |
| 15 | 15 |
| 16 #include "webrtc/base/safe_conversions.h" | 16 #include "webrtc/base/safe_conversions.h" |
| 17 #include "webrtc/common_audio/signal_processing/include/signal_processing_librar
y.h" | 17 #include "webrtc/common_audio/signal_processing/include/signal_processing_librar
y.h" |
| 18 #include "webrtc/modules/audio_coding/neteq/background_noise.h" | 18 #include "webrtc/modules/audio_coding/neteq/background_noise.h" |
| 19 #include "webrtc/modules/audio_coding/neteq/cross_correlation.h" | |
| 20 #include "webrtc/modules/audio_coding/neteq/dsp_helper.h" | 19 #include "webrtc/modules/audio_coding/neteq/dsp_helper.h" |
| 21 | 20 |
| 22 namespace webrtc { | 21 namespace webrtc { |
| 23 | 22 |
| 24 TimeStretch::ReturnCodes TimeStretch::Process(const int16_t* input, | 23 TimeStretch::ReturnCodes TimeStretch::Process(const int16_t* input, |
| 25 size_t input_len, | 24 size_t input_len, |
| 26 bool fast_mode, | 25 bool fast_mode, |
| 27 AudioMultiVector* output, | 26 AudioMultiVector* output, |
| 28 size_t* length_change_samples) { | 27 size_t* length_change_samples) { |
| 29 // Pre-calculate common multiplication with |fs_mult_|. | 28 // Pre-calculate common multiplication with |fs_mult_|. |
| (...skipping 122 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
| 152 break; | 151 break; |
| 153 case kNoStretch: | 152 case kNoStretch: |
| 154 case kError: | 153 case kError: |
| 155 *length_change_samples = 0; | 154 *length_change_samples = 0; |
| 156 break; | 155 break; |
| 157 } | 156 } |
| 158 return return_value; | 157 return return_value; |
| 159 } | 158 } |
| 160 | 159 |
| 161 void TimeStretch::AutoCorrelation() { | 160 void TimeStretch::AutoCorrelation() { |
| 161 // Set scaling factor for cross correlation to protect against overflow. |
| 162 int scaling = kLogCorrelationLen - WebRtcSpl_NormW32( |
| 163 max_input_value_ * max_input_value_); |
| 164 scaling = std::max(0, scaling); |
| 165 |
| 162 // Calculate correlation from lag kMinLag to lag kMaxLag in 4 kHz domain. | 166 // Calculate correlation from lag kMinLag to lag kMaxLag in 4 kHz domain. |
| 163 int32_t auto_corr[kCorrelationLen]; | 167 int32_t auto_corr[kCorrelationLen]; |
| 164 CrossCorrelationWithAutoShift( | 168 WebRtcSpl_CrossCorrelation(auto_corr, &downsampled_input_[kMaxLag], |
| 165 &downsampled_input_[kMaxLag], &downsampled_input_[kMaxLag - kMinLag], | 169 &downsampled_input_[kMaxLag - kMinLag], |
| 166 kCorrelationLen, kMaxLag - kMinLag, -1, auto_corr); | 170 kCorrelationLen, kMaxLag - kMinLag, scaling, -1); |
| 167 | 171 |
| 168 // Normalize correlation to 14 bits and write to |auto_correlation_|. | 172 // Normalize correlation to 14 bits and write to |auto_correlation_|. |
| 169 int32_t max_corr = WebRtcSpl_MaxAbsValueW32(auto_corr, kCorrelationLen); | 173 int32_t max_corr = WebRtcSpl_MaxAbsValueW32(auto_corr, kCorrelationLen); |
| 170 int scaling = std::max(0, 17 - WebRtcSpl_NormW32(max_corr)); | 174 scaling = std::max(0, 17 - WebRtcSpl_NormW32(max_corr)); |
| 171 WebRtcSpl_VectorBitShiftW32ToW16(auto_correlation_, kCorrelationLen, | 175 WebRtcSpl_VectorBitShiftW32ToW16(auto_correlation_, kCorrelationLen, |
| 172 auto_corr, scaling); | 176 auto_corr, scaling); |
| 173 } | 177 } |
| 174 | 178 |
| 175 bool TimeStretch::SpeechDetection(int32_t vec1_energy, int32_t vec2_energy, | 179 bool TimeStretch::SpeechDetection(int32_t vec1_energy, int32_t vec2_energy, |
| 176 size_t peak_index, int scaling) const { | 180 size_t peak_index, int scaling) const { |
| 177 // Check if the signal seems to be active speech or not (simple VAD). | 181 // Check if the signal seems to be active speech or not (simple VAD). |
| 178 // If (vec1_energy + vec2_energy) / (2 * peak_index) <= | 182 // If (vec1_energy + vec2_energy) / (2 * peak_index) <= |
| 179 // 8 * background_noise_energy, then we say that the signal contains no | 183 // 8 * background_noise_energy, then we say that the signal contains no |
| 180 // active speech. | 184 // active speech. |
| (...skipping 23 matching lines...) Expand all Loading... |
| 204 int temp_scale = WebRtcSpl_NormW32(left_side); | 208 int temp_scale = WebRtcSpl_NormW32(left_side); |
| 205 left_side = left_side << temp_scale; | 209 left_side = left_side << temp_scale; |
| 206 right_side = right_side >> (2 * scaling - temp_scale); | 210 right_side = right_side >> (2 * scaling - temp_scale); |
| 207 } else { | 211 } else { |
| 208 left_side = left_side << 2 * scaling; | 212 left_side = left_side << 2 * scaling; |
| 209 } | 213 } |
| 210 return left_side > right_side; | 214 return left_side > right_side; |
| 211 } | 215 } |
| 212 | 216 |
| 213 } // namespace webrtc | 217 } // namespace webrtc |
| OLD | NEW |