Study On Robot Acoustic Source Localization System Based On Microphone Array

The development of artificial intelligent voice interaction technology is changing rapidly,and gradually aroused people's widespread concern.Robot auditory system as a kind of natural,convenient and efficient,intelligent way of robot to interact with the external system,in the process of obtaining speech signals,indoor environment noise and reverberation,and the interference of other speakers is inevitable.In order to solve the problems of low positioning accuracy,poor real-time performance and high cost of acoustic source positioning system in the complex environment with noise and reverberation,this paper carry out a further study of robot acoustic source localization based on microphone array,and the specific works were summarized as follows:(1)Discussed the far-field model and the near-field model of signal propagation models,and considering the distribution of noise and reverberation in indoor environment,established an array signal receiving model single-source multi-reverb model based on noise and reverberation environment.Introduced microphone array bracket design technology,and according to the design principle of array structure and the actual demands for acoustic localization system,a quaternary cross-shaped microphone array structure was designed,and its geometrical space positioning method was deduced.(2)Introduced the characteristics of speech and space sampling theorem of array signals.Because of the acquisition process of speech signals in the indoor environment,it inevitably suffered from the interference of environment noise and reverberation.Therefore,it highlights the effects of pre-processing and voice activity detection.It introduces the pretreatment operations such as filter de-noising and windows,and introduces the voice activity detection to avoid the wrong location,and to improve the running efficiency of the whole system.(3)Analyzed the advantages and disadvantages of three traditional acoustic source localization methods,and it focused on the several kinds of commonly used positioning methods based on time delay estimation,including the generalized cross-correlation time delay estimation algorithm and LMS adaptive time delay estimation algorithm.On this basis,it introduced the quadratic correlation method for improving GCC time-delay estimation algorithm,and put forward an improved sound source location algorithm based on delay estimation.The experimental results show that the proposed algorithm effectivelyimprove the precision of time delay estimation,its performance is superior in strong noise and reverberation environment.(4)Established a feasible real-time acoustic source positioning system,and introduced the platform composition of the system in detail.Introduced the hardware structure of microphone array and action execution module in detail,and the software design flow of the system is given.By three groups of experiment data,it analyzed the sampling frequency,the array element spacing and the distance from the sound source to the array center on the result of positioning.The experimental results show that the higher the sampling frequency and the greater the array element spacing,the sound source localization results are more accurate,and the location results are influenced by the distance from the sound source to the array center farther or nearer,the proper sound source distance will bring good positioning effect.Finally,the experiment verifies that the performance of the improved algorithm and the feasibility of the sound source localization system,and summed up the factors affected the positioning accuracy,including the limitation of the algorithm itself,the hardware factors and environmental factors,etc.