A connected digit speech recognition is important in many applications such as voice-dialing telephone, automated banking system, automatic data entry, PIN entry, etc. This research presents speech recognition system of speaker-independent Thai connected digit. The system employs mel frequency cepstrum coefficient (MFCC), delta MFCC, delta-delta MFCC, delta energy and delta-delta energy as features, and applies continuous density hidden Markov model (CDHMM) in the recognition process. The Viterbi beam search algorithm is used in decoding process. In training set, we use 100 speakers (50 females, 50 males) for 2000 utterances within the range of 20-28 years old. For the experiment, we used 50 speakers (25 females, 25 males) as testing set. The average recognition rate is 75.25 % for known length strings and 70.33 % for unknown length strings.
Cite as: Deemagarn, A., Kawtrakul, A. (2004) Thai connected digit speech recognition using hidden Markov models. Proc. 9th Conference on Speech and Computer (SPECOM 2004), 731-735
@inproceedings{deemagarn04_specom,
author={Amarin Deemagarn and Asanee Kawtrakul},
title={{Thai connected digit speech recognition using hidden Markov models}},
year=2004,
booktitle={Proc. 9th Conference on Speech and Computer (SPECOM 2004)},
pages={731--735}
}