Speech Masking Based On Artificial Synthesis

In today's era of rapid development of information science and technology,intelligent devices emerge one after another,communication devices such as computers and smart phones are constantly updated,and public social networking apps are increasing.People begin to pay attention to their privacy.People's daily communication means is language,through which information can be transmitted to each other,including a large number of private information,especially in the national major conference,the voice contains a large number of state secrets.China's comprehensive national strength continues to increase,some foreign intelligence agencies in order to obtain our country's secrets,through a variety of ways to carry out secret theft activities,to the country's political field,economic field,national defense deployment and other security poses a certain threat.Confidential meeting rooms and other places of classified mainly used for national meeting or used to communicate important science and technology,the content of the conversation here can't leak,sound protection is required,the current means of voice protection is a physical protection,the combined laser eavesdropping and sound masking systems,this article is for sound masking systems were studied.Acoustic masking system is to use a sound to protect the target speech,mainly to study the masking source in the system,two new masking sources were applied to the system and the masking effect was tested.The main work contents are as follows:First,the masking sound of synthetic speech.The current masking sound cannot guarantee good masking effect and auditory comfort at the same time.Therefore,a new masking sound based on artificial speech synthesis is proposed.The masking sound has a very high correlation with the target speech,and it is difficult for eavesdroppers to recognize the voice when it is acquired.In the test,different synthetic masking sounds were obtained by adjusting the duration and fundamental frequency to find the best masking sound.The results show that,compared with the existing masking sound,this masking sound has better masking effect and good auditory comfort.The synthetic masking sound solves the problem that masking effect and comfort can not be taken into account simultaneously.Second,mixed source masking sound.In order to solve the problem of single sound source and unadjustable sound pressure in acoustic masking system,a hybrid sound source with adjustable sound pressure was proposed.The sound pressure regulation function of the masked sound source is to obtain the sound pressure of the target speech through Ensemble Empirical Mode Decomposition(EEMD)and A-weight fitting,so that the sound pressure difference between the masked sound and the target speech remains unchanged.Because the mixing of various interference noises will result in poor auditory comfort,two or three kinds of interference noises are selected as the mixed sound source,including masking sound of synthesized speech.The experimental results show that the masking effect of single masking sounds decreases gradually when the sound pressure of target speech changes,while the masking effect of mixed masking sounds always keeps good.