OLD | NEW |
| (Empty) |
1 /* | |
2 * Copyright (c) 2012 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 WEBRTC_MODULES_AUDIO_PROCESSING_VAD_VAD_AUDIO_PROC_H_ | |
12 #define WEBRTC_MODULES_AUDIO_PROCESSING_VAD_VAD_AUDIO_PROC_H_ | |
13 | |
14 #include "webrtc/base/scoped_ptr.h" | |
15 #include "webrtc/modules/audio_processing/vad/common.h" | |
16 #include "webrtc/typedefs.h" | |
17 | |
18 namespace webrtc { | |
19 | |
20 class AudioFrame; | |
21 class PoleZeroFilter; | |
22 | |
23 class VadAudioProc { | |
24 public: | |
25 // Forward declare iSAC structs. | |
26 struct PitchAnalysisStruct; | |
27 struct PreFiltBankstr; | |
28 | |
29 VadAudioProc(); | |
30 ~VadAudioProc(); | |
31 | |
32 int ExtractFeatures(const int16_t* audio_frame, | |
33 int length, | |
34 AudioFeatures* audio_features); | |
35 | |
36 static const int kDftSize = 512; | |
37 | |
38 private: | |
39 void PitchAnalysis(double* pitch_gains, double* pitch_lags_hz, int length); | |
40 void SubframeCorrelation(double* corr, int length_corr, int subframe_index); | |
41 void GetLpcPolynomials(double* lpc, int length_lpc); | |
42 void FindFirstSpectralPeaks(double* f_peak, int length_f_peak); | |
43 void Rms(double* rms, int length_rms); | |
44 void ResetBuffer(); | |
45 | |
46 // To compute spectral peak we perform LPC analysis to get spectral envelope. | |
47 // For every 30 ms we compute 3 spectral peak there for 3 LPC analysis. | |
48 // LPC is computed over 15 ms of windowed audio. For every 10 ms sub-frame | |
49 // we need 5 ms of past signal to create the input of LPC analysis. | |
50 static const int kNumPastSignalSamples = kSampleRateHz / 200; | |
51 | |
52 // TODO(turajs): maybe defining this at a higher level (maybe enum) so that | |
53 // all the code recognize it as "no-error." | |
54 static const int kNoError = 0; | |
55 | |
56 static const int kNum10msSubframes = 3; | |
57 static const int kNumSubframeSamples = kSampleRateHz / 100; | |
58 static const int kNumSamplesToProcess = | |
59 kNum10msSubframes * | |
60 kNumSubframeSamples; // Samples in 30 ms @ given sampling rate. | |
61 static const int kBufferLength = kNumPastSignalSamples + kNumSamplesToProcess; | |
62 static const int kIpLength = kDftSize >> 1; | |
63 static const int kWLength = kDftSize >> 1; | |
64 | |
65 static const int kLpcOrder = 16; | |
66 | |
67 int ip_[kIpLength]; | |
68 float w_fft_[kWLength]; | |
69 | |
70 // A buffer of 5 ms (past audio) + 30 ms (one iSAC frame ). | |
71 float audio_buffer_[kBufferLength]; | |
72 int num_buffer_samples_; | |
73 | |
74 double log_old_gain_; | |
75 double old_lag_; | |
76 | |
77 rtc::scoped_ptr<PitchAnalysisStruct> pitch_analysis_handle_; | |
78 rtc::scoped_ptr<PreFiltBankstr> pre_filter_handle_; | |
79 rtc::scoped_ptr<PoleZeroFilter> high_pass_filter_; | |
80 }; | |
81 | |
82 } // namespace webrtc | |
83 | |
84 #endif // WEBRTC_MODULES_AUDIO_PROCESSING_VAD_VAD_AUDIO_PROC_H_ | |
OLD | NEW |