中文口語處理技術之前瞻性研究課題( 1/3 )

With the rapid developments of wireless communications, it is highly desired for users to access the network
information with spoken dialogue interface via hand-held devices at any time, from anywhere. One possible
approach towards this goal is to perform speech feature extraction at the hand-held devices (the clients) and have
all other recognition tasks and dialogue functions absorbed by the server. This report investigated distributed
Chinese keyword spotting and verification under this scenario. A “phonetically distributed” Mandarin speech
database including all possible Mandarin syllables and context relationships with frequencies roughly proportional
to those occurring in daily Mandarin conversation is used to train a best set of vector quantization codebooks, such
that the syllable recognition accuracy degradation due to quantization errors is minimized. Enhanced Chinese
keyword spotting techniques were then developed using utterance verification approaches with weighting
parameters optimized by MCE training. Experimental results indicated that the keyword verification approach
achieved significant improvements in keyword spotting performance, and the overall results integrating vector
quantization, keyword spotting and verification is quite satisfactory.