We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
A COMPARATIVE SURVEY OF DTW AND HMM USING HAUSA ISOLATED DIGITS RECOGNITION IN HUMAN COMPUTER INTERACTION SYSTEM.
- Authors
Ibrahim, Yakubu A.; Ibiyemi, Tunji S.
- Abstract
Speech Recognition is a vital part of different computer-based applications in communication and security systems. However, there has been very little research in the aspect of speech Human Computer Interaction system for African languages such as Hausa, hence, the need to extend the research in order to bring in, the different systems based on speech recognition. Also, Hausa is an important ethnic tribe of lingua franca in both west and central Africa countries. Isolated word recognition is an easy speech type because it demands the user to pause between each word. In this study, the two algorithms that were used to implement a system of Recognition of Hausa isolated digits are Dynamic Time Warping and Hidden Markov Model. To perform the recognition efficiently, speech endpoint, framing blocking, speech normalization, vector quantization and Mel Frequency Cepstral Coefficient techniques were used to process the speech. The accuracy of about 94% was obtained for recognition with HMM-based system. In a very noisy environment, the performance of the two techniques is bad but the pattern matching using HMM is better than the pattern matching using DTW.
- Subjects
HUMAN-computer interaction; AUTOMATIC speech recognition; HAUSA language; WORD recognition; HIDDEN Markov models
- Publication
Annals. Computer Science Series, 2017, Vol 15, Issue 2, p61
- ISSN
1583-7165
- Publication type
Article