speech_probability_buffer.h (2951B)
1 /* 2 * Copyright (c) 2022 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_SPEECH_PROBABILITY_BUFFER_H_ 12 #define MODULES_AUDIO_PROCESSING_AGC2_SPEECH_PROBABILITY_BUFFER_H_ 13 14 #include <vector> 15 16 #include "rtc_base/gtest_prod_util.h" 17 18 namespace webrtc { 19 20 // This class implements a circular buffer that stores speech probabilities 21 // for a speech segment and estimates speech activity for that segment. 22 class SpeechProbabilityBuffer { 23 public: 24 // Ctor. The value of `low_probability_threshold` is required to be on the 25 // range [0.0f, 1.0f]. 26 explicit SpeechProbabilityBuffer(float low_probability_threshold); 27 ~SpeechProbabilityBuffer() {} 28 SpeechProbabilityBuffer(const SpeechProbabilityBuffer&) = delete; 29 SpeechProbabilityBuffer& operator=(const SpeechProbabilityBuffer&) = delete; 30 31 // Adds `probability` in the buffer and computes an updatds sum of the buffer 32 // probabilities. Value of `probability` is required to be on the range 33 // [0.0f, 1.0f]. 34 void Update(float probability); 35 36 // Resets the histogram, forgets the past. 37 void Reset(); 38 39 // Returns true if the segment is active (a long enough segment with an 40 // average speech probability above `low_probability_threshold`). 41 bool IsActiveSegment() const; 42 43 private: 44 void RemoveTransient(); 45 46 // Use only for testing. 47 float GetSumProbabilities() const { return sum_probabilities_; } 48 49 FRIEND_TEST_ALL_PREFIXES(SpeechProbabilityBufferTest, 50 CheckSumAfterInitialization); 51 FRIEND_TEST_ALL_PREFIXES(SpeechProbabilityBufferTest, CheckSumAfterUpdate); 52 FRIEND_TEST_ALL_PREFIXES(SpeechProbabilityBufferTest, CheckSumAfterReset); 53 FRIEND_TEST_ALL_PREFIXES(SpeechProbabilityBufferTest, 54 CheckSumAfterTransientNotRemoved); 55 FRIEND_TEST_ALL_PREFIXES(SpeechProbabilityBufferTest, 56 CheckSumAfterTransientRemoved); 57 58 const float low_probability_threshold_; 59 60 // Sum of probabilities stored in `probabilities_`. Must be updated if 61 // `probabilities_` is updated. 62 float sum_probabilities_ = 0.0f; 63 64 // Circular buffer for probabilities. 65 std::vector<float> probabilities_; 66 67 // Current index of the circular buffer, where the newest data will be written 68 // to, therefore, pointing to the oldest data if buffer is full. 69 int buffer_index_ = 0; 70 71 // Indicates if the buffer is full and adding a new value removes the oldest 72 // value. 73 int buffer_is_full_ = false; 74 75 int num_high_probability_observations_ = 0; 76 }; 77 78 } // namespace webrtc 79 80 #endif // MODULES_AUDIO_PROCESSING_AGC2_SPEECH_PROBABILITY_BUFFER_H_