Automatic Speaker Recognition System in Urdu using MFCC & HMM
Shaik Riyaz1, Bathula Lakshmi Bhavani2, S. Venkatrama Phani Kumar4
1Shaik Riyaz, Vignan’s Foundation for Science, Technology & Research, (A.P), India.
2Bathula Lakshmi Bhavani, Vignan’s Foundation for Science, Technology & Research, (A.P), India.
3S.Venkatrama Phani Kumar, Vignan’s Foundation for Science, Technology & Research, (A.P), India.
Manuscript received on 12 February 2019 | Revised Manuscript received on 02 March 2019 | Manuscript Published on 08 June 2019 | PP: 109-113 | Volume-7 Issue-5S4, February 2019 | Retrieval Number: E10210275S419/19©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Speech is one of the most common ways of communication between users and it is also serves to recognize the individual. In this paper, an automatic speaker recognition system with Mel-Frequency Cepstral Coefficients (MFCC) and Hidden Markov Model (HMM) is proposed to recognize the identity of the users using Urdu utterances. MFCC is a very popular feature extraction approach to extract features with human auditory behavior. In the view of feature size and to increase the efficiency, acoustic precise feature extraction is carried with Vector quantization (VQ). HMM will make the recognition process simple and much more realistic. Performance of the proposed model is evaluated on a dataset with 250 isolated Urdu words uttered by twenty speakers, out of which eight speakers are male and twelve speakers are female. The proposed model outperforms with 96.4% of accuracy when compared with other models.
Keywords: Hidden Markov Model (HMM), Mel-Frequency Cepstral Coefficients (MFCC), Vector Quantization.
Scope of the Article: Pattern Recognition