Isolated Odia Digit Recognition Using HTK: An Implementation View

This paper outlines the frame work for an isolated Odia digit recognition. The HTK toolkit which consider the Hidden Markov Model (HMM), as the base for developing the system. The recording environment is simple and created using the HTK tool. Initially the system is trained for 900 signals of 10 Odia digits spoken by 10 different speakers. The work done on this paper also signifies the role of HTK tool commands, applied in various levels of the system advancement. 10 samples, (each sample consists of 10 untrained signal) from the untrained 100 signals are considered for testing at each time. The trial results show that the average performance of our system is encouraging.