DEVELOPING THE IDENTIFICATION SYSTEM FOR ANURAN VOCALIZATIONS

作者：Huang Chenn Jung^*; Liao Jia Jian; Hsu Po An; Yang Yi Ju; Yang Dian Xiu; Chen You Jia

来源：International Journal of Innovative Computing Information and Control, 2010, 6(10): 4421-4433.

• Data mining
• Pattein recognition
• Anuran detection
• Outlier detection
• Wrapper feature selection

An intelligent identification system for anuran vocalizations is developed in this work to provide the public to easily consult online The raw anuran vocalization samples air first filtered by noise removal. high frequency compensation, and discrete wavelet. transform techniques in outer An adaptive end-point. detection segmentation algorithm is pm posed to effectively separate the individual syllables from the noise Seven features. including spectral centroid, signal bandwidth, spectral roll-off, threshold-crossing rate, delta Spectrum magnitude, spectral flatness, (and average energy, are extracted and serve as the input pour actors of wrapper feature selection method Selected features air used to assisting in separating the anuran calls from non-anuran sounds. Then k-nearest neighbor classifier (kNN) and Gaussian mixture model are used to classify the any ran species based on another chosen feature set A series of experiments woe conducted to measure the outcome performance of the pin posed work. Experimental insults exhibit. that the recognition rate of k-nearest neighbor classifier using GMM non-anuran detection method with two different feature sets can achieve up to 80 42%. The effectiveness of the proposed identification system for anuran vocalizations that uses the combination with anuran detection and classification is thus verified.

• 出版日期2010-10
• 单位东华大学