Reserch Of Robot Apeech Recognition And Soung Source Localization In The Disaster Site Rescue

With the advancement of technology, research in the field of disaster environmentwhich robots can replace human work has practical significance worthy of attention. Theauditory system, as an important part, has brought more and more attention.The speechrecognition and the sound source localization are two of the important main component ofapeech signal processing. At the same time there are many cross-section between the twostudies with high research value. Sound source localization with a high applicability, easyto implement and low cost. With these advantages makes sound source localization hasbecome a research focus. More with a good prospect.This paper is supported by the Hi-tech Research and Development Program of China--Adaptive disaster site information gathering ground robot(No.2012AA041508). It makes aspecific in-depth research into speech recognition and sound source localization.This paper covers the following research points:1. Speech recognition has been researched. Pre-processing the received speech signal.The most suitable window function has been got through experiments. Achieved the soundendpoint detection through double threshold comparison method, identified sound start andend points. To overcome the problem of sound source localization invalid.2. Hardware and software systems of the whole system has been researched. Throughresearch on the principle of selection by the microphone and microphone array designrequirements. Combined with near-field and far-field microphone division model.Ultimately selected DELL four microphone with a font array.Using NI USB-9234dataacquisition card connected with the host computer equipped with a microphone connecteddevices. Software section is based on MATLAB and LabVIEW programming languagemix, give full play to their respective advantages.3. A sound source localization method of the microphone array is researched. Acomparative analysis of the experimental on three kinds of sound source localizationmethods. Analyzed the generalized cross-correlation delay estimation algorithm. Andcarried out a detailed comparison experiment of the weighting function which is the maininfluencing factor. Different weighting function in in different SNR obtain different positioning results by MATLAB Simulation. The appropriate method has been finalchoiced. Greatly improve the accuracy of time delay estimation positioning algorithm.4. By using software programming and hardware device conduct test-lab environmentexperiment. In the form of graphs and charts show the experimental results. The experimentresults show that the experiment achieved the objective of the speech recognition and thesound source localization in the laboratory environment. Then the error of the experimentalresults were analyzed to identify deficiencies, and the need for future research andimprovements made suggestions. Laid the foundation for future research.