THE DYNAMICALLY-ADJUSTABLE HISTOGRAM PRUNING METHOD FOR EMBEDDED VOICE DIALING

摘要

Memory and speed are two key factors that must be faced when applying voice dialer to Pocket PCs. To provide a solution, a novel decoding method integrated with the score differences of token paths is proposed, named as "Dynamically-Adjustable Histogram Pruning". Additionally, the computation of likelihood score is accelerated by means of dynamic score lookup table. Furthermore, a new acoustic modeling method based on Extended Initial/Final (XIF) with less dimensioned acoustic feature is proven suitable for embedded speech command recognition. By adopting the methods developed above, we implement a speaker-independent, user definable voice dialing speech recognition system with good performance on a real PDA device. For a 200-Chinese-word vocabulary, its recognition accuracy reaches 97.80%. Meanwhile, it obtains better recognition speed by 80 times and saves decoding space by 30% in comparison to the baseline system using standard Viterbi decoding method.