To search, Click below search items. |
| All Published Papers Search Service |
|
|
|
Title |
Text-Dependent Speaker Recognition System for Indian Languages |
|
Author |
R. Rajeswara Rao, A. Nagesh, Kamakshi Prasad, K. Ephraim Babu |
| Citation |
Vol. 7 No. 11 pp. 65-71 |
|
Abstract |
Speaker Recognition is a process of automatically recognising who is speaking on the basis of speaker dependent features of the speech signal. Although speaker recognition is currently not as robust as other biometrics such as finger prints and retinal scans, speech holds great promise. Speech based recognition permits remote access. Speech is very important in a country like India with a large population with low levels of literacy and education. Speech empowers people by helping them to overcome the barriers of language and complexity of usage. In this paper we describe a system for speaker recognition designed with low security access control systems in mind. An isolated word speech recognition system is used to recognize the spoken password and then a speaker identification system is used to further confirm the identity of the user amongst a given set of users. Mel Frequency Cepstral Coefficients have been used to build Hidden Markov Models. The HTK tool kit has been used to build these systems. Mahalanobis distance measure is employed. Experiments conducted are described and the results are shown. Mostly spoken ten Indian languages speech data has been used in these experiments. It can be seen that the system gives very good performance for the intended task. |
|
Keywords |
Hidden Markov Models, Speaker Recognition, HTK |
|
URL |
