Research On The Remote Laser Coherent Speech Signal Detection Techonology

Compared with traditional speech detection technology,laser speech detection can detect speech signal in long-distance and non-contact situations,and has the advantages of strong concealment and convenient operation.Therefore,it is widely used in military,security,surveillance and other fields.Common laser speech detection technologies(such as light reflection modulation measurement technology and speckle image measurement technology)have shortcomings such as poor environmental applicability and complicated adjustment of measurement equipment,which cannot meet the practical application requirements of laser speech detection.Laser coherent Doppler vibration measurement technology has the advantages of long distance,high precision,non-contact,high spatial resolution and wide dynamic range.It has become the main detection method in the field of laser speech detection.Foreign related systems have achieved 300 meters of speech detection,meanwhile a series of related researches have been carried out in China.However,compared with foreign countries,there is still a certain gap in terms of the distance of operation(under the same laser power)and the quality of speech signal acquisition.In order to achieve long-distance,high-quality speech detection in China,more in-depth researches are needed in terms of optical path structure design,measurement target selection,and speech signal processing.Therefore,this paper focuses on the above three aspects,using theoretical analysis,numerical simulation and experimental research combined method to carry out in-depth research on long-range laser coherent speech signal detection technology.The main contents of this paper are as follows:(1)This paper introduces the development status of laser speech detection technology at home and abroad.Characteristics of common laser speech detection methods are analyzed and compared.Additionally,the method about applying the laser coherent Doppler vibration measurement technology into laser speech detection system is proposed.The development history of laser coherent Doppler vibration measurement technology is reviewed and the research status at home and abroad is introduced.Besides,the technical difficulties that need to be solved in the development of laser coherent speech detection technology in China are summarized,which provides guidance for further in-depth research.(2)The principle of speech signal detection based on laser coherent Doppler vibration measurement technology is expounded.Mathematical expressions for assessment criteria which are used to evaluate the performance of the system such as the carrier-to-noise ratio,minimum resolution and detection probability of the laser coherent speech detection system are given by theoretical calculation.In addition,through theoretical derivation and numerical analysis,the effects of optical factors such as aberration,laser line width,detection mode and emission crosstalk on the performance of laser coherent speech detection system are studied.Finally,based on the optical path structure of the all-fiber laser coherent voice detection system,partial-fiber structures are designed and built to solve the problem of insufficient isolation of the fiber circulator.In addition,the optical components of the system are determined based on the results of the theoretical research.(3)The essence of laser coherent speech detection is to detect the vibration of an object caused by sound pressure.Therefore,the selection of the object to be measured directly affects the quality of the detected speech signal.According to the STL theory and the SSC model,the vibration model of the flat plates with different materials and shapes in the sound field is used to simulate the vibration amplitude of the object in the sound field,and the accuracy of the analysis is verified by experiments.During laser speech detection,objects near the sound source are rough compared to the wavelength of the light.Therefore,this paper discusses the influence of different rough surfaces on the carrier-to-noise ratio of laser coherent speech detection system using the combination of numerical simulation and experiments.When the rough surface has motion other than the optical axis direction relative to the speech detection system,speckle noise will be introduced to the speech detection process.In this paper,the influence of speckle noise on the detected speech is discussed based on the speckle characteristics,and the experimental verification is carried out.In addition,the power spectral density of speckle noise is calculated theoretically.The above research can provide guidance for selecting suitable measurement targets during the detection process.(4)Several common demodulation methods are introduced systematically.Aiming at the requirement that the laser interception system needs to ensure the accuracy and real-time during the signal processing at the same time,the demodulation algorithm combined with orthogonal demodulation and arctangent detection is used to reconstruct the speech signal.Due to environmental noise,rough surfaces of objects,etc.,reconstructed speech is subject to noise interference.Therefore,the noise reduction processing is required.According to the representation of noise in coherently detected speech signal,the noise can be divided into background noise and impact noise.For background noise,this paper uses the method combined with Optimal Logarithmic Spectrum estimation algorithm(OM-LSA)and phase compensation to reduce noise;For the impact noise,this paper first uses the kurtosis ratio to determine the position of the noise in the speech signal,and then uses the linear prediction method to estimate the speech data contaminated by the impact noise.In addition,since the vibration response of the measured object to the sound field will cause the bandwidth of the coherent detection speech to be narrow,this paper uses the half-wave rectification-based speech bandwidth extension algorithm to extend the coherent detection speech bandwidth.(5)Speech detection experiments were conducted and the indoor distances are 30 m,70m,120 m,while the vibration measurement targets are folder(plastic),beverage bottles(paper)and computer display.The experimental results show that the self-developed system can effectively acquire long-distance(at least 120m)understandable speech signals.On the basis of fully demonstrating the performance of the self-research detection system,the dual-mode monitoring experiment and the long-distance speaker recognition experiment were carried out respectively.The results of the two kinds of experiments show the feasibility and potential of the self-developed speech detection system in the field of security monitoring and biometric identification.In summary,this paper has gradually solved various key difficulties through the theoretical and experimental research on the key technologies of long-range laser coherent speech signal detection,and laid the foundation for the subsequent realization of longer-distance and higher-quality speech detection.