Non-uniform unit parsing for SSS-LR continuous speech recognition

Author

Singer, Harald ; Takami, Jun-Ichi ; Matsunaga, Shoichi

Author_Institution

ATR Interpreting Telecommun.Res. Labs., Kyoto, Japan

fYear

1994

fDate

19-22 Apr 1994

Abstract

We describe recent improvements in ATR´s experimental speech recognition system ATREUS, which serves as a recognition font end for the speech translation system ASURA. Our next goal is spontaneous speech translation. To constrain the potentially huge search space, better prosodic control, better probabilistic language models and better acoustic models are proposed. The SSS-LR parser was modified to work with non-uniform unit type acoustic and duration models. Experimental results showed, that, for example, use of mora trigram probabilities improved the phrase error rate from 17% to 14%

Keywords

context-free grammars; language translation; natural languages; probability; speech recognition; speech recognition equipment; ASURA; ATR; ATREUS; CGF grammar; SSS-LR continuous speech recognition; acoustic models; duration models; experimental results; mora trigram probabilities; nonuniform unit type acoustic models; phrase error rate; probabilistic language models; prosodic control; recognition font end; speech recognition system; speech translation system; successive state splitting; Acoustic beams; Context modeling; Databases; Hidden Markov models; Natural languages; Speech recognition; Speech synthesis; Topology; Training data;

fLanguage

English

Publisher

ieee

Conference_Titel

Acoustics, Speech, and Signal Processing, 1994. ICASSP-94., 1994 IEEE International Conference on

Conference_Location

Adelaide, SA

ISSN

1520-6149

Print_ISBN

0-7803-1775-0

Type

conf

DOI

10.1109/ICASSP.1994.389697

Filename

389697