Improvement Of The Text Dependent Speaker Identification System Using Discrete MMM With Cepstral Based Features

Daffodil International University Journal of Science and Technology

Muhammad Abdul Goffar Khan¹  Md Rabiul Islam¹  Department of Electrical & Electronic Engineering, Rajshahi University of Engineering & Technology (RUET), RajshahiBangladesh¹  Md Fayzur Rahman¹ 

关键词: Biometric Technologies;    Automatic Speaker Identification;    Cepstral Coefficients;    Feature Extraction;    Hidden Markov Model.;   

DOI  :  10.3329/diujst.v6i2.9341

学科分类：自然科学（综合）

来源: Daffodil International University

【 摘 要 】

In this paper, an improved strategy for automated text based speaker identification scheme has been proposed. The identification process incorporates the Hidden Markov Model technique. After preprocessing the speech, HMM is used in the learning and identification. Features are extracted by different techniques such as RCC, MFCC, ΔMFCC, ΔΔMFCC, LPC and LPCC which is almost different in each case. The highest identification rate of 93% has been achieved in the close set text dependent speaker identification system. Keywords: Biometric Technologies; Automatic Speaker Identification; Cepstral Coefficients; Feature Extraction; Hidden Markov Model.

【 授权许可】

Unknown