spectral_features.h (3254B)
1 /* 2 * Copyright (c) 2018 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 #ifndef MODULES_AUDIO_PROCESSING_AGC2_RNN_VAD_SPECTRAL_FEATURES_H_ 12 #define MODULES_AUDIO_PROCESSING_AGC2_RNN_VAD_SPECTRAL_FEATURES_H_ 13 14 #include <array> 15 #include <cstddef> 16 #include <memory> 17 18 #include "api/array_view.h" 19 #include "modules/audio_processing/agc2/rnn_vad/common.h" 20 #include "modules/audio_processing/agc2/rnn_vad/ring_buffer.h" 21 #include "modules/audio_processing/agc2/rnn_vad/spectral_features_internal.h" 22 #include "modules/audio_processing/agc2/rnn_vad/symmetric_matrix_buffer.h" 23 #include "modules/audio_processing/utility/pffft_wrapper.h" 24 25 namespace webrtc { 26 namespace rnn_vad { 27 28 // Class to compute spectral features. 29 class SpectralFeaturesExtractor { 30 public: 31 SpectralFeaturesExtractor(); 32 SpectralFeaturesExtractor(const SpectralFeaturesExtractor&) = delete; 33 SpectralFeaturesExtractor& operator=(const SpectralFeaturesExtractor&) = 34 delete; 35 ~SpectralFeaturesExtractor(); 36 // Resets the internal state of the feature extractor. 37 void Reset(); 38 // Analyzes a pair of reference and lagged frames from the pitch buffer, 39 // detects silence and computes features. If silence is detected, the output 40 // is neither computed nor written. 41 bool CheckSilenceComputeFeatures( 42 ArrayView<const float, kFrameSize20ms24kHz> reference_frame, 43 ArrayView<const float, kFrameSize20ms24kHz> lagged_frame, 44 ArrayView<float, kNumBands - kNumLowerBands> higher_bands_cepstrum, 45 ArrayView<float, kNumLowerBands> average, 46 ArrayView<float, kNumLowerBands> first_derivative, 47 ArrayView<float, kNumLowerBands> second_derivative, 48 ArrayView<float, kNumLowerBands> bands_cross_corr, 49 float* variability); 50 51 private: 52 void ComputeAvgAndDerivatives( 53 ArrayView<float, kNumLowerBands> average, 54 ArrayView<float, kNumLowerBands> first_derivative, 55 ArrayView<float, kNumLowerBands> second_derivative) const; 56 void ComputeNormalizedCepstralCorrelation( 57 ArrayView<float, kNumLowerBands> bands_cross_corr); 58 float ComputeVariability() const; 59 60 const std::array<float, kFrameSize20ms24kHz / 2> half_window_; 61 Pffft fft_; 62 std::unique_ptr<Pffft::FloatBuffer> fft_buffer_; 63 std::unique_ptr<Pffft::FloatBuffer> reference_frame_fft_; 64 std::unique_ptr<Pffft::FloatBuffer> lagged_frame_fft_; 65 SpectralCorrelator spectral_correlator_; 66 std::array<float, kOpusBands24kHz> reference_frame_bands_energy_; 67 std::array<float, kOpusBands24kHz> lagged_frame_bands_energy_; 68 std::array<float, kOpusBands24kHz> bands_cross_corr_; 69 const std::array<float, kNumBands * kNumBands> dct_table_; 70 RingBuffer<float, kNumBands, kCepstralCoeffsHistorySize> 71 cepstral_coeffs_ring_buf_; 72 SymmetricMatrixBuffer<float, kCepstralCoeffsHistorySize> cepstral_diffs_buf_; 73 }; 74 75 } // namespace rnn_vad 76 } // namespace webrtc 77 78 #endif // MODULES_AUDIO_PROCESSING_AGC2_RNN_VAD_SPECTRAL_FEATURES_H_