Advances in Arabic broadcast news transcription at RWTH

Author

Rybach, David ; Hahn, Stefan ; Gollan, Christian ; Schlüter, Ralf ; Ney, Hermann

Author_Institution

RWTH Aachen Univ., Aachen

fYear

2007

fDate

9-13 Dec. 2007

Firstpage

449

Lastpage

454

Abstract

This paper describes the RWTH speech recognition system for Arabic. Several design aspects of the system, including cross-adaptation, multiple system design and combination, are analyzed. We summarize the semi-automatic lexicon generation for Arabic using a statistical approach to grapheme-to-phoneme conversion and pronunciation statistics. Furthermore, a novel ASR-based audio segmentation algorithm is presented. Finally, we discuss practical approaches for parallelized acoustic training and memory efficient lattice rescoring. Systematic results are reported on recent GALE evaluation corpora.

Keywords

audio signal processing; broadcasting; natural language processing; speech recognition; statistical analysis; Arabic broadcast news transcription; GALE evaluation corpora; RWTH speech recognition system; audio segmentation algorithm; global autonomous language exploitation; grapheme-to-phoneme conversion; memory efficient lattice rescoring; parallelized acoustic training; pronunciation statistics; semi automatic lexicon generation; Broadcasting; Cepstral analysis; Hidden Markov models; Humans; Lattices; Loudspeakers; Mel frequency cepstral coefficient; Natural languages; Neural networks; Speech recognition; Audio Segmentation; Cross-Adaptation; Speech Recognition; System Combination;

fLanguage

English

Publisher

ieee

Conference_Titel

Automatic Speech Recognition & Understanding, 2007. ASRU. IEEE Workshop on

Conference_Location

Kyoto

Print_ISBN

978-1-4244-1746-9

Electronic_ISBN

978-1-4244-1746-9

Type

conf

DOI

10.1109/ASRU.2007.4430154

Filename

4430154