The Implementation And Optimization Of Sound Recovery Based On Subtle Vibration Extraction Algorithm

Human voice is transmitted through the form of sound waves.Sound waves can cause the vibration of the air and then lead to the vibration of human hearing organs.The voice produced by human vocal organs is also called voice which often contains certain information and represents different phonetic meanings.However,the speech caused by acoustic vibration is usually very weak,it needs to be amplified and recovered to be heard.People often use microphone to amplify the vibration through the circuit.In recent years,the form of optical microphone has also appeared,and the amplification and recovery of speech can be realized by laser reflection.Visual measurement is a rapidly improved measurement technology in recent years.It has been widely used in the fields of shape detection,target recognition and tracking and object structure and motion analysis.It has many advantages,such as non-contact,full field measurement and so on.High-speed vision measurement,as a way of visual measurement,can get better frequency information,which is suitable for occasions where precise measurement is needed.Considering the vibration produced by speech is also a kind of small motion,this paper is concentrated on the realization of extraction of subtle vibration and sound recovery based on high-speed visual measurement.This paper studies the method of subtle vibration extraction based on high-speed visual measurement.First,it analyzes three different image magnification techniques,compares their advantages and disadvantages,and finds the algorithm based on phase amplification,which is suitable for motion amplification.The related simulation experiment is conducted,too,to verify the effect of the algorithm.Then it proposed an improved subtle vibration extraction method based on singular value decomposition,it optimize the selection of the sub region.By using the characteristic of singular value decomposition,the expression of the essential feature of motion characteristics can be found.For the realization of sound recovery,the whole system architecture is designed,the key parts of the system are introduced.Then the basic experiment is carried out on the basis of the system,which realizes the amplification and recovery of the speech information.Under the condition that final result is not very ideal for the direct results,a spectral subtraction based on noise estimation is proposed to optimize and enhance the speech.Compared to the traditional spectral subtraction,it achieves a more accurate estimation and separation of noise,and it finally rechieves a good result.In this paper,an improved subtle vibration extraction algorithm based on singular value decomposition is proposed.It can realize the recovery of sound and have a good effect.It has a practical significance in the fields of security,monitoring and so on.At the same time,subtle vibration also contains many other signals except the sound,so the extraction and denoising method also has some reference for the amplification and recovery of the similar vibration signals.