Hidden Information Transmission Based On Breath Sounds Demarcation Algorithm In Instant Call

Breathing is the symbol of vitality.As an inevitable physiological activity to maintain life,the importance of breathing is self-evident.But for a long time,in speech signal processing,researchers mostly pay their attention to the part of voiced sounds.Because of breath sounds' low energy and low frequency,they are usually treated as a negligible part or even noise.However,as a sound signal that is significantly different from the voiced sounds,breath sounds themselves can also reflect a lot of useful information.Recently,more and more researchers have begun to notice the connotation behind breath sounds.They began to analyze just only breath sounds,including its use in the medical field,emotion recognition field and speaker recognition field.This paper intends to explore another use of breath sounds.Because it's hard to notice the existence of breath sounds during speech,so we can achieve the purpose of covert information transmission by giving different meanings of different breath sounds.Firstly,this paper defines the new concept of breathing gesture.A breathing gesture is a set of well-designed breath sounds with intrinsic meanings.Users embed secret information into speech by inserting breathing gestures during a call.As a carrier carrying secret information,the design of breathing gestures is an important part of this paper's work.In order to achieve the purpose of covert information transmission,the breathing gesture should satisfy the following characteristics: diversity,conciseness,concealment and distinction with other breath sounds.Based on these,this paper designs three different breathing gestures and analyzes and tests them separately.The extraction of secret information is implemented by a set of algorithms.The algorithm id divided into four steps: Firstly,pre-processing the original signal to suppress the noise;Then using a template matching method based on MFCC to remove the voiced part of the original call;The reserved breath sounds are further processed,and several isolated breathing gestures are obtained by segmentation;Finally,these isolated breathing gestures are recognized.The recognition algorithm is a classification based on DTW distance,and the GFCC is used as the feature of breathing gestures.Finally,in order to evaluate the performance of the whole system,the algorithm of extraction of breathing gesture and the algorithm of the recognition of breathing gesture are tested respectively.The experimental results show that recognition rate of breathing gesture can reach 87.6%.