Speaker-independent isolated word recognition using word-based vector quantization and hidden Markov models

In this paper, we investigate the possibility of using word-based vector quantization with hidden Markov models for speaker-independent isolated word recognition. Two word-based algorithms were proposed and studied. Experiments were carried out on Chinese (Cantonese) digits spoken by 110 speakers (55 males and 55 females) in two databases. An improvement of about 3% in recognition rate was obtained in one of the word-based algorithms. The results and implications are discussed.