DocumentCode
575181
Title
Clustering and modeling of hazardous or non-hazardous sound in daily life
Author
Suzuki, Ken ; Tanaka, Motoshi ; Inoue, Hiroshi
Author_Institution
Dept. of Electr. & Electron. Eng., Akita Univ., Akita, Japan
fYear
2012
fDate
20-23 Aug. 2012
Firstpage
1991
Lastpage
1995
Abstract
A system which detects hazardous sounds and situations in daily life was proposed. Feature vectors were consisted of spectral flux and 12 MFCCs which were extracted from a recorded daily sound. Clustering of daily sound (non-hazardous sound) by the complete link method was discussed. Probability model of the daily sound was created, and ten simulated hazardous sounds were examined with the probability model. Feasibility of detecting hazardous sound by the probability model was confirmed.
Keywords
cepstral analysis; pattern clustering; probability; MFCC; Mel-frequency cepstrum coefficients; clustering method; complete link method; daily life; daily sound; feature vectors; hazardous sound; nonhazardous sound; probability model; spectral flux; Accidents; Educational institutions; Feature extraction; Microphones; Probability; Senior citizens; Vectors; clustering; daily sound; hazardous sound; probability modeling;
fLanguage
English
Publisher
ieee
Conference_Titel
SICE Annual Conference (SICE), 2012 Proceedings of
Conference_Location
Akita
ISSN
pending
Print_ISBN
978-1-4673-2259-1
Type
conf
Filename
6318339
Link To Document