DocumentCode :
2541085
Title :
Unsupervised Speaker Clustering Using SVM Training Missclassification Rate for Short-Duration Speech Signals
Author :
Lin, Po-Chuan ; Jui, Yeh-Yi ; Ying, Tsai-Sung ; Chen, Yeong-Chin ; Wu, Menq-Jion
Author_Institution :
Dept. of Electron. Eng. & Comput. Sci., Tung-Fang Design Univ., Kaohsiung, Taiwan
fYear :
2010
fDate :
13-15 Dec. 2010
Firstpage :
606
Lastpage :
609
Abstract :
This paper proposes an unsupervised speaker clustering system for duration of speech signals below 4 seconds. For determining whether two collected speech sections uttered from the same speaker or not, our previous SVM training miss-classification rate (STMR) is adopted to evaluate the data separability between two different speakers. This paper also proposes a hierarchical extract and merge (HEM) clustering method to reduce agglomeration time and enhance the clustering purity. Experiment results show the average speaker purity (ASP) and average cluster purity (ACP) are both better than the CE manner with the GMM training miss-classification rates (GTMR) for 2 to 4 seconds short speech sections.
Keywords :
Gaussian processes; pattern clustering; speaker recognition; speech processing; support vector machines; GMM training misclassification rates; SVM training misclassification rate; agglomeration time; average cluster purity; average speaker purity; data separability; hierarchical extract and merge clustering; short-duration speech signals; unsupervised speaker clustering; Acoustics; Classification algorithms; Clustering algorithms; Hidden Markov models; Speech; Support vector machines; Training; SVM Training Miss-classification Rate (STMR); Speaker Clustering; Support Vector Machine (SVM);
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Genetic and Evolutionary Computing (ICGEC), 2010 Fourth International Conference on
Conference_Location :
Shenzhen
Print_ISBN :
978-1-4244-8891-9
Electronic_ISBN :
978-0-7695-4281-2
Type :
conf
DOI :
10.1109/ICGEC.2010.155
Filename :
5715505
Link To Document :
بازگشت