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_AGC_STANDALONE_VAD_H_ 12 #define WEBRTC_MODULES_AUDIO_PROCESSING_AGC_STANDALONE_VAD_H_ 13 14 #include "webrtc/base/scoped_ptr.h" 15 #include "webrtc/modules/audio_processing/vad/common.h" 16 #include "webrtc/common_audio/vad/include/webrtc_vad.h" 17 #include "webrtc/typedefs.h" 18 19 namespace webrtc { 20 21 class AudioFrame; 22 23 class StandaloneVad { 24 public: 25 static StandaloneVad* Create(); 26 ~StandaloneVad(); 27 28 // Outputs 29 // p: a buffer where probabilities are written to. 30 // length_p: number of elements of |p|. 31 // 32 // return value: 33 // -1: if no audio is stored or VAD returns error. 34 // 0: in success. 35 // In case of error the content of |activity| is unchanged. 36 // 37 // Note that due to a high false-positive (VAD decision is active while the 38 // processed audio is just background noise) rate, stand-alone VAD is used as 39 // a one-sided indicator. The activity probability is 0.5 if the frame is 40 // classified as active, and the probability is 0.01 if the audio is 41 // classified as passive. In this way, when probabilities are combined, the 42 // effect of the stand-alone VAD is neutral if the input is classified as 43 // active. 44 int GetActivity(double* p, size_t length_p); 45 46 // Expecting 10 ms of 16 kHz audio to be pushed in. 47 int AddAudio(const int16_t* data, size_t length); 48 49 // Set aggressiveness of VAD, 0 is the least aggressive and 3 is the most 50 // aggressive mode. Returns -1 if the input is less than 0 or larger than 3, 51 // otherwise 0 is returned. 52 int set_mode(int mode); 53 // Get the agressiveness of the current VAD. 54 int mode() const { return mode_; } 55 56 private: 57 explicit StandaloneVad(VadInst* vad); 58 59 static const size_t kMaxNum10msFrames = 3; 60 61 // TODO(turajs): Is there a way to use scoped-pointer here? 62 VadInst* vad_; 63 int16_t buffer_[kMaxNum10msFrames * kLength10Ms]; 64 size_t index_; 65 int mode_; 66 }; 67 68 } // namespace webrtc 69 70 #endif // WEBRTC_MODULES_AUDIO_PROCESSING_AGC_STANDALONE_VAD_H_ 71
