Open AccessResearch on Crane Sound Clustering of MFCC Based on HHT
Author(s) -
Qinjian Fu,
Danjue Lv,
Yan Zhang,
Haifeng Xu,
Yao Wang,
Jiali Zi,
Jiang Liu
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1693/1/012134
Subject(s) - mel frequency cepstrum , hilbert–huang transform , cluster analysis , speech recognition , pattern recognition (psychology) , feature (linguistics) , computer science , filter (signal processing) , feature extraction , artificial intelligence , computer vision , linguistics , philosophy
Due to the uniqueness of the sound mechanism of birds, they have typical non-stationary and nonlinear characteristics. This paper proposed a new acousic feature, HHT-MFCC, combined the HHT transformation and MFCC method, aiming at the dynamic instantaneousness of bird sounds. This method, firstly, uses the ensemble empirical mode decomposition EEMD to decompose the bird sounds into a number of intrinsic modal functions IMFs, and then adopts the Hilbert transform to obtain the Hilbert marginal spectrum of each IMF, at last applies the mel-scale filter to complete the feature extraction of HHT-MFCC. The experiment extracts HHT-MFCC from 9 kinds of cranes in China to cluster. Three indexes of cluster evaluation are used to evaluate the feautre HHT-MFCC and MFCC. The results show that the HHT-MFCC feature is 10% higher in RI index than MFCC, 9% higher in JC index, and 4% higher in FMI index.