next up previous
次へ: Beam search 上へ: A Spontaneous Speech Recognition 戻る: Recognition Algorithm Using Word

Memory and Computational Cost Reduction


This algorithm is of complexity $O ( Q^2 \cdot
l_w \cdot T ) $ as opposed to $ O( Q
\cdot l_w \cdot T ) $ for a bigram grammar. This entails a large memory and huge computational cost. To overcome these problems, the following methods are adopted.




Subsections

Jin'ichi Murakami 平成13年1月19日