Kriengkrai Nantanitikron. A prototype of Thai speech recognition system for basic voice commanding . Master's Degree(Technology of Information System Management). Mahidol University. : Mahidol University, 2008.
A prototype of Thai speech recognition system for basic voice commanding
Abstract:
Most recent Thai speech recognition systems that have been continually researched often use feature extraction from Fourier transform based methods and some learning algorithm such as Artificial Neural Networks (ANN), Hidden Markov Models (HMM), and Linear Prediction Codes (LPC). This study presents a Thai speech recognition system based on Fourier transform and a set of filter banks as a feature extraction which is called double filter banks. The goal of this study was to develop a prototype of a real-time single-word Thai speech recognition system that can recognize some common Thai words. The study encompasses speech capturing, analog-to-digital conversion, automatic speech marking, identification of starting and ending points, pre-emphasis, speech feature extraction, speech feature matching, and displaying the output to user. The author creates an application to analyze a comprehensive set of speech data and created a multi-dimensional speech feature that includes speech achieving operations such as read, write, and load into the main memory in JAVA language. The system was evaluated for its accuracy and stability in performing various conditions. The accuracy was validated by an experiment with 9,000 speeches from several volunteers. The average accuracy rate is 94.6% in an offline test. Finally, the result shows that the evaluation was beyond satisfaction for every aspect