Comparative Study of Continuous Hidden Markov Models (CHMM) and Artificial Neural Network (ANN) on Speaker Identification System

This paper reports a comparative study between continuous hidden Markov model (CHMM) and artificial neural network (ANN) on text dependent, closed set speaker identification (SID) system with Thai language recording in office environment. Thai isolated digit 0-9 and their concatenation are used as speaking text. Mel frequency cepstral coefficients (MFCC) are selected as the studied features. Two well-known recognition engines, ANN and CHMM, are conducted and compared. The ANN system (multilayer perceptron network with backpropagation learning algorithm) is applied with a special design of input feeding methods. The general Gaussian density distribution HMM is developed for CHMM system. After optimizing system’s parameters by performing some preliminary experiments, CHMM gives the best identification rate at 90.4%, which is slightly better than 90.1% of ANN on digit “5”. Contrarily when using 3-concatenated digit, ANN achieves 97.3%, which is higher than 95.7% of CHMM. Keyword: Speaker identification (SID), Thai language, Continuous hidden Markov model (CHMM), Artificial neural network (ANN)