Robust speech recognition in burst-like packet loss

Author

Milner, Ben

Author_Institution

BT Adastral Park, Martlesham Heath, UK

Volume

1

fYear

2001

fDate

2001

Firstpage

261

Abstract

This paper examines problems associated with performing speech recognition over mobile and IP networks. The main problems are identified as codec-based distortion and from speech vectors being lost from packet loss in the network. A realistic model for packet loss is developed, based on a three state Markov model and is shown to be capable of simulating the burst-like nature of packet loss. A two stage packet loss detection and estimation scheme is proposed and is shown to improve the recognition performance in the event of feature vectors being lost. Results from the Aurora database show that burst-like packet loss reduces the digit accuracy from 99% to 57% at 50% packet loss. Estimation of the lost packets recovers the performance to 77%

Keywords

buried object detection; hidden Markov models; land mobile radio; packet radio networks; parameter estimation; speech codecs; speech recognition; transport protocols; Aurora database; HMM; IP networks; burst-like packet loss; codec-based distortion; digit accuracy; mobile networks; packet loss model; recognition performance; robust speech recognition; speech vectors; three state Markov model; two stage packet loss detection; two stage packet loss estimation; Acoustic distortion; Code standards; IP networks; Robustness; Spatial databases; Speech analysis; Speech codecs; Speech processing; Speech recognition; Telecommunication standards;

fLanguage

English

Publisher

ieee

Conference_Titel

Acoustics, Speech, and Signal Processing, 2001. Proceedings. (ICASSP '01). 2001 IEEE International Conference on

Conference_Location

Salt Lake City, UT

ISSN

1520-6149

Print_ISBN

0-7803-7041-4

Type

conf

DOI

10.1109/ICASSP.2001.940817

Filename

940817