next up previous
次へ: Experimental conditions for continuous 上へ: main1 戻る: Entropy for the Ergodic

Experiments for Continuous Speech Recognition

To show the effectiveness of the Ergodic HMM as a language model, speaker dependent continuous speech recognition experiments were carried out. In these experiments, a standard speech recognition algorithm employed a continuous mixture HMM and beam search, with the Ergodic HMM providing the language information[6].



Subsections

Jin'ichi Murakami 平成13年10月2日