DocumentCode :
2261033
Title :
A new search algorithm in segmentation lattices of speech signals
Author :
Husson, Jean-Luc ; Laprie, Yves
Author_Institution :
CRIN-INRIA, CNRS, Vandoeuvre-les-Nancy, France
Volume :
4
fYear :
1996
fDate :
3-6 Oct 1996
Firstpage :
2099
Abstract :
This paper describes a new segmentation system using a multi-level representation called a dendrogram. We address the issues of estimating the confidence of one path and finding the N most reliable paths in the segmentation lattice. Our approach rests on automatically trained criteria and on an efficient strategy to prune the search space
Keywords :
reliability; signal representation; speech processing; speech recognition; tree searching; automatically trained criteria; dendrogram; most reliable paths; multi-level representation; path confidence estimation; search algorithm; search space pruning strategy; segmentation lattices; speech signals; Acoustic measurements; Algorithm design and analysis; Automatic speech recognition; Lattices; Mel frequency cepstral coefficient; Random variables; Robustness; Speech analysis; Speech recognition;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Spoken Language, 1996. ICSLP 96. Proceedings., Fourth International Conference on
Conference_Location :
Philadelphia, PA
Print_ISBN :
0-7803-3555-4
Type :
conf
DOI :
10.1109/ICSLP.1996.607216
Filename :
607216
Link To Document :
بازگشت