Research On Time Delay Estimation Method Based On Microphone Array

Compared with a single microphone signal processing, microphone array can provide more information on airspace, and with more advantages in terms of enhancing the signal and determines the acoustic source location, so it has a wide range of applications in real life. TDE technology based on microphone array is mainly used in the field of acoustic source localization, it is also an indispensable important link in acoustic source localization technology based on microphone array. Acoustic source localization based on microphone arrays refers to that the location of acoustic source is determined by the signal information received from several microphones arrayed with a certain geometric structure, then processing the received acoustic source signal information by microphone array, resulting acoustic source signal arrival time delay of the microphone, according to the acoustic source arrival time delay between the two microphones and the geometric structure of microphone array to determine the location of the acoustic source. Estimated time delay is an important part influencing positioning. This paper focuses on the time-delay estimation techniques based on microphone array technology in the field of acoustic source localization technology.Aiming at noise has greater impact on microphone array for source location accuracy, the paper focuses on the sound source signal de-noising technology, and combines de-noising technique with calculation time delay algorithm, reducing noise interference to the accuracy of time delay.The noise signal should be preprocessed before the calculation time delay and effective noise reduction will directly affect the accuracy of positioning. In order to improve the de-noising effect, the basic principle of wavelet threshold de-noising was studied, on the basis of the research of traditional threshold function, proposing an improved wavelet threshold function. On the basis of soft threshold function, the adjustment factor is introduced to structure the improved threshold function, according to the function curve of the improved threshold function to analyze the de-noising performance. The theoretical analysis and simulation results show that the improved function ameliorated the problems that constant bias exists in soft threshold function and hard threshold function discontinuous, which effectively filter out noise in speech signal, and it retains the characteristic information of the original speech signal in a greater degree.As the noise has a bad impact on the precision of time delay estimation, the wavelet threshold function de-noising was combined with the generalized cross-correlation algorithm. The time delay estimation of the noise signal was processed by using the time delay estimation method based on the wavelet threshold function. The simulation results show that the cross-correlation function which was processed by using the generalized cross-correlation method based on the wavelet threshold function, it presents the characteristics of small fluctuations and sharp peak, the results of time delay are accurate.In terms of array layout, the paper mainly studied the positioning principle and algorithm of the four-element cross array. Combined with time delay estimation method it based on improved wavelet threshold function to do sound source location positioning experiment. The experimental results show that the locating method has a good performance and can be considered to use in practice.