5th European Conference on Speech Communication and Technology

Rhodes, Greece
September 22-25, 1997

New Methods in Continuous Mandarin Speech Recognition

C. Julian Chen (1), Ramesh A. Gopinath (1), Michael D. Monkowski (1), Michael A. Picheny (1), Katherine Shen (2)

(1) IBM Thomas J. Watson Research Center, Yorktown Heights, NY, USA (2) IBM China Research Laboratory, Beijing , China

We describe new methods for speaker-independent, continuous mandarin speech recognition based on the IBM HMM-based continuous speech recognition system (1-3): First, we treat tones in mandarin as attributes of certain phonemes, instead of syllables. Second, instantaneous pitch is treated as a variable in the acoustic feature vector, in the same way as cepstra or energy. Third, by designing a set of word-segmentation rules to convert the continuous Chinese text into segmented text, an effective trigram language model is trained(4). By applying those new methods, a speaker-independent, very-large-vocabulary continuous mandarin dictation system is demonstrated. Decoding results showed that its performance is similar to the best results for US English.

Full Paper

Bibliographic reference.  Chen, C. Julian / Gopinath, Ramesh A. / Monkowski, Michael D. / Picheny, Michael A. / Shen, Katherine (1997): "New methods in continuous Mandarin speech recognition", In EUROSPEECH-1997, 1543-1546.