A Research Based On The Technique Of Unit Identification Of The Animal Sound

With the number of wild animals has steadily declined, people begin to place paramount importance on the protection of them and research on them is also increasing. Therefore, the research on animal acoustics has drawn people's attention. Taxonomists are taking the sound signal of animals as a standard of classification of animal through investigation;ecologists are probing into how to reveal the mystery of animal action by sound signal. To be more noticeable, animal sound can be used to direct animals and make them do as human being's command.This paper presents a thorough study of the end point detection of Energy Frequency Value, LPCC, MFCC and their difference characteristic parameter, the recognition methods of applying VQ and HMM to sound recognition by studying the basic elements and systematic structure of sound recognition and analyzing the present technique of sound recognition. Thus, the system of individual identification based on animal sound is established.This paper obtains the characteristic parameter of sound input by MATLAB sound processing toolbox, in which LPCC, MFCC and their difference are taken as the characteristic parameter of sound signal, introduced LBG algorithm going on yards of book design, Viterbi algorithm is adopted to do training and identification. A higher discrimination is achieved. And further more it has the advantages of faster operating speed, little calculation amount, lower error rate and so on.This paper makes a basic and preparatory study for exploring practical sound recognition system further by testing and investigating actual sound recognition system.