Bird Species Recognition by Wavelet Transformation of a Section of Birdsong

Author

Chou, Chih-Hsun ; Liu, Pang-Hsin

Author_Institution

Dept. of Comput. Sci. & Inf. Eng., Chung Hua Univ., Hsinchu, Taiwan

fYear

2009

fDate

7-9 July 2009

Firstpage

189

Lastpage

193

Abstract

In this study, a method for birdsong recognition is proposed. In this method, after detecting the range of each syllable, birdsong sections containing a period of syllables were segmented. For each syllable of a birdsong section, the first five orders MFCCs were computed, and the same order MFCCs of all syllables were aligned so that wavelet transformation can be applied to compute the feature vector of the birdsong section. By using neural network as the classifier, the proposed method was applied to recognize the birdsongs of 420 bird species.

Keywords

acoustic signal processing; neural nets; wavelet transforms; zoology; MFCC; bird species recognition; birdsong recognition; neural network; wavelet transformation; Birds; Computer science; Conferences; Feature extraction; Histograms; Mel frequency cepstral coefficient; Neural networks; Pervasive computing; Power harmonic filters; Prototypes; Birdsong recognition; Mel-frequency cepstral coefficients; syllable segmentation; wavelet transformation;

fLanguage

English

Publisher

ieee

Conference_Titel

Ubiquitous, Autonomic and Trusted Computing, 2009. UIC-ATC '09. Symposia and Workshops on

Conference_Location

Brisbane, QLD

Print_ISBN

978-1-4244-4902-6

Electronic_ISBN

978-0-7695-3737-5

Type

conf

DOI

10.1109/UIC-ATC.2009.85

Filename

5319240