Phonetic Modelling in the Philips Chinese Continuous Speech Recognition System

We have extended the Philips large vocabulary continuous speech recognition system towards Chinese On the way from our existing Western language technology to Mandarin the rst step was to build a suitable phonetic model This paper describes the development of our phonetic model excluding tones for Mandarin Chinese We will present a systematic comparison of three forms of sub syllabic units for Chinese phonemes initials nals and a non tonal form of preme toneme models as well as whole syllable models for reference We include experiments on bottom up and decision tree based top down state clustering and modelling of cross syllable contexts All forms of sub syllabic units are represented in the Philips Mandarin phone set SAMPA C SAMPA C is based on the European SAMPA standard and introduced in this paper Our studies show that traditional half syllable approaches slightly outperform Western style triphones Modelling of right context dependency gives greater improvement than left context dependency and cross syllable modelling yields a performance gain In a free syllable decoding task we achieve syllable error rate for telephone speech and for microphone dictations