Sub-sentence discourse models for conversational speech recognition

Author

Ma, Kristine W. ; Zavaliagkos, George ; Meteer, Marie

Author_Institution

GTE/BBN Technol., Cambridge, MA, USA

Volume

2

fYear

1998

fDate

12-15 May 1998

Firstpage

693

Abstract

According to discourse theories in linguistics, conversational utterances possess an informational structure that partitions each sentence into two portions: a “given” and “new”. We explore this idea by building sub-sentence discourse language models for conversational speech recognition. The internal sentence structure is captured in statistical language modeling by training multiple n-gram models using the expectation-maximization algorithm on the Switchboard corpus. The resulting model contributes to a 30% reduction in language model perplexity and a small gain in word error rate

Keywords

grammars; natural languages; speech processing; speech recognition; statistical analysis; Switchboard corpus; conversational speech recognition; conversational utterances; expectation-maximization algorithm; given-new language model; informational structure; internal sentence structure; language model perplexity reduction; linguistics; multiple n-gram models training; statistical language modeling; sub-sentence discourse models; word error rate; Acoustic testing; Acoustic waves; Automatic testing; Data analysis; Data mining; Error analysis; Performance gain; Speech recognition; Training data;

fLanguage

English

Publisher

ieee

Conference_Titel

Acoustics, Speech and Signal Processing, 1998. Proceedings of the 1998 IEEE International Conference on

Conference_Location

Seattle, WA

ISSN

1520-6149

Print_ISBN

0-7803-4428-6

Type

conf

DOI

10.1109/ICASSP.1998.675359

Filename

675359