Thursday 25 April 2013

Sistem Pengecaman Huruf Aksara Menggunakan Teknik Dynamic Time Warping


Abstract

          The subject of speech recognition has been a topic of research for four decades. With the advent of new hardware technology and software algorithm, techniques for speech recognition have evolved. The aim of speech recognition is to make computers understand what you speak. If it were totally enabled, you might think you need not a keyboard for your computer, and it is helpful especially for those who have difficulties in using keyboards including handicapped people. However, the cormercial speech recognition system still can't be at human brain performance . This is because of disadvantages of the hardware such as microphones, or the noises or the recognition techniques itselfs. That is why so many researches is in speech recognition carry on until now to overcome the problems that occur. In this project, Dynamic Time Warping is choosen to be the recognition techniques. The purpose of DTW is to produce a warping function that minimizes the total distance between the respective points of the signals. The method used the template based approach. Distance measures are used to compare templates to find the best match. Hence, Dynamic programming is used to solve temporal differerence. The study of this method only using Itakura global constraint. From the research that has been made using 20 different individuals as a testers for the system, percentage of the result is 70.5%. This value can be improved in future research to produce an accurate, robust and reliable speech recognition system.

No comments:

Post a Comment