standalone_vad.h (2321B)
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 MODULES_AUDIO_PROCESSING_AGC_STANDALONE_VAD_H_ 12 #define MODULES_AUDIO_PROCESSING_AGC_STANDALONE_VAD_H_ 13 14 #include <stddef.h> 15 #include <stdint.h> 16 17 #include "common_audio/vad/include/webrtc_vad.h" 18 #include "modules/audio_processing/vad/common.h" 19 20 namespace webrtc { 21 22 class StandaloneVad { 23 public: 24 static StandaloneVad* Create(); 25 ~StandaloneVad(); 26 27 // Outputs 28 // p: a buffer where probabilities are written to. 29 // length_p: number of elements of `p`. 30 // 31 // return value: 32 // -1: if no audio is stored or VAD returns error. 33 // 0: in success. 34 // In case of error the content of `activity` is unchanged. 35 // 36 // Note that due to a high false-positive (VAD decision is active while the 37 // processed audio is just background noise) rate, stand-alone VAD is used as 38 // a one-sided indicator. The activity probability is 0.5 if the frame is 39 // classified as active, and the probability is 0.01 if the audio is 40 // classified as passive. In this way, when probabilities are combined, the 41 // effect of the stand-alone VAD is neutral if the input is classified as 42 // active. 43 int GetActivity(double* p, size_t length_p); 44 45 // Expecting 10 ms of 16 kHz audio to be pushed in. 46 int AddAudio(const int16_t* data, size_t length); 47 48 // Set aggressiveness of VAD, 0 is the least aggressive and 3 is the most 49 // aggressive mode. Returns -1 if the input is less than 0 or larger than 3, 50 // otherwise 0 is returned. 51 int set_mode(int mode); 52 // Get the agressiveness of the current VAD. 53 int mode() const { return mode_; } 54 55 private: 56 explicit StandaloneVad(VadInst* vad); 57 58 static const size_t kMaxNum10msFrames = 3; 59 60 // TODO(turajs): Is there a way to use scoped-pointer here? 61 VadInst* vad_; 62 int16_t buffer_[kMaxNum10msFrames * kLength10Ms]; 63 size_t index_; 64 int mode_; 65 }; 66 67 } // namespace webrtc 68 69 #endif // MODULES_AUDIO_PROCESSING_AGC_STANDALONE_VAD_H_