Introduction of a working method for biological signals classification. Application to bird chirp recognition

We present a working method for solving die problem of automatic acoustic s ignature recognition and illustrate by the classification of a bird chirp. The first step of the process is time and frequency analysis. This primordi al stage allows us to define the acquisition chain caracteristics. Moreover , it sheds ligth on the choices that are to be made during the next stages. The second step, the choice of the modelling technique, should enable us t o extract a set of discriminant parameters from the signal, with respect to the classification considered. Finally, the third step consists of choosin g a classification method suited to our problem and testing its performance .