Morpheme-Based Language Modeling for Arabic Lvcsr

Author

Choueiter, Ghinwa ; Povey, Daniel ; Chen, Stanley F. ; Zweig, Geoffrey

Author_Institution

MIT CS, Cambridge, MA

Volume

1

fYear

2006

fDate

14-19 May 2006

Abstract

In this paper, we concentrate on Arabic speech recognition. Taking advantage of the rich morphological structure of the language, we use morpheme-based language modeling to improve the word error rate. We propose a simple constraining method to rid the decoding output of illegal morpheme sequences. We report the results obtained for word and morpheme language models using medium (64 kw) and large (~800 kw) vocabularies, the morpheme LM obtaining an absolute improvement of 2.4% for the former and only 0.2% for the latter. The 2.4% gain surpasses previous gains for morpheme-based LMs for Arabic, and the large vocabulary runs represent the first comparative results for vocabularies of this size for any language. Finally, we analyze the performance of the morpheme LM on word OOV´s

Keywords

decoding; natural languages; speech coding; speech recognition; Arabic LVCSR; Arabic speech recognition; constraining method; decoding output; morpheme sequences; morpheme-based language modeling; word error rate; Artificial intelligence; Decoding; Error analysis; Laboratories; Natural languages; Paints; Performance analysis; Speech recognition; Testing; Vocabulary;

fLanguage

English

Publisher

ieee

Conference_Titel

Acoustics, Speech and Signal Processing, 2006. ICASSP 2006 Proceedings. 2006 IEEE International Conference on

Conference_Location

Toulouse

ISSN

1520-6149

Print_ISBN

1-4244-0469-X

Type

conf

DOI

10.1109/ICASSP.2006.1660205

Filename

1660205