Title :
Speaker age recognition based on isolated words by using SVM
Author :
Mengdi Yue ; Ling Chen ; Jie Zhang ; Hong Liu
Author_Institution :
Sch. of Electron. & Inf. Eng., Univ. of Sci. & Technol. Liaoning, Anshan, China
Abstract :
Speaker age recognition is an essential technique in automation speech recognition based on the speech wavform parameters in speaker´s voice. However, there are several challenges in speaker age recognition, such as innate differences in speaker´s voice, subjective classification fuzzy, etc. The issue of speaker age based on isolated words is proposed in this paper, including support vector machine (SVM), picking up Mel frequency cepstrum coefficient (MFCC) characteristics of isolated words to distinguish the speaker´s age. The voicebox of this paper includes 4507 isolated word speech. Experimental results show that the recognition rate based on isolated word speech can reach 72.93%. Through the experiment towards SVM classifier, we could find that the performance is improved without normalization for MFCC.
Keywords :
age issues; speech recognition; support vector machines; Mel frequency cepstrum coefficient; SVM; automation speech recognition; isolated words; speaker age recognition; speech wavform parameters; support vector machine; voicebox; Acoustics; Polynomials; Speech; Speech recognition; Support vector machines; Testing; Training; Age recognition; Isolated words; Mel frequency cepstrum coefficients; Support vector machine;
Conference_Titel :
Cloud Computing and Intelligence Systems (CCIS), 2014 IEEE 3rd International Conference on
Print_ISBN :
978-1-4799-4720-1
DOI :
10.1109/CCIS.2014.7175743
