DocumentCode
3210600
Title
Speech segmentation in working environments
Author
Chen, Xian ; Brown, Peter ; Bovey, John
Author_Institution
Comput. Lab., Kent Univ., Canterbury, UK
fYear
1997
fDate
35557
Firstpage
42522
Lastpage
42525
Abstract
A speech segmentation method has been developed for working environments. This segmentation is designed to be supervised, i.e. pre-captured speech samples of potential speakers are used to train the system because we need to know who the speakers are. In order to be usable in working environments a number of measures are taken to increase the correctness of identification. The segmentation can be used as an audio indexing system to provide a quick access to audio materials in memory-aid systems
Keywords
speaker recognition; Gaussian model; audio indexing system; audio material access; memory-aid systems; office automation; speaker identification; speech recognition; speech samples; speech segmentation; supervised segmentation; training; working environments;
fLanguage
English
Publisher
iet
Conference_Titel
Prospects for Spoken Language Technology (Digest No: 1997/138), IEE Colloquium on
Conference_Location
London
Type
conf
DOI
10.1049/ic:19970764
Filename
643194
Link To Document