Publication Details

Score Normalization and System Combination for Improved Keyword Spotting

KARAKOS Damianos, SCHWARTZ Richard, TSAKALIDIS Stavros, ZHANG Le, RANJAN Shivesh, NG Tim, HSIAO Roger, NGUYEN Long, GRÉZL František, HANNEMANN Mirko, KARAFIÁT Martin, SZŐKE Igor and VESELÝ Karel et al. Score Normalization and System Combination for Improved Keyword Spotting. In: Proceedings of ASRU 2013. Olomouc: IEEE Signal Processing Society, 2013, pp. 210-215. ISBN 978-1-4799-2755-5. Available from: https://ieeexplore.ieee.org/document/6707731
Czech title
Normalizace skóre a kombinace systémů pro vylepšenou detekci klíčových slov
Type
conference paper
Language
english
Authors
Karakos Damianos (Raytheon BBN)
Schwartz Richard (Raytheon BBN)
Tsakalidis Stavros (Raytheon BBN)
Zhang Le (Raytheon BBN)
Ranjan Shivesh (Raytheon BBN)
Ng Tim (Raytheon BBN)
Hsiao Roger (Raytheon BBN)
Nguyen Long (Raytheon BBN)
Grézl František, Ing., Ph.D. (DCGM FIT BUT)
Hannemann Mirko, Dipl.-Ing. (DCGM FIT BUT)
Karafiát Martin, Ing., Ph.D. (DCGM FIT BUT)
Szőke Igor, Ing., Ph.D. (DCGM FIT BUT)
and others
URL
Keywords

keyword search, score normalization, system combination, indexing and search

Abstract

This article is about Score Normalization and System Combination for Improved Keyword Spotting (KWS).

Annotation

We present two techniques that are shown to yield improved Key- word Spotting (KWS) performance when using the ATWV/MTWV performance measures: (i) score normalization, where the scores of different keywords become commensurate with each other and they more closely correspond to the probability of being correct than raw posteriors; and (ii) system combination, where the detections of multiple systems are merged together, and their scores are in- terpolated with weights which are optimized using MTWV as the maximization criterion. Both score normalization and system com- bination approaches show that significant gains in ATWV/MTWV can be obtained, sometimes on the order of 8-10 points (absolute), in five different languages. A variant of these methods resulted in the highest performance for the official surprise language evaluation for the IARPA-funded Babel project in April 2013.

Published
2013
Pages
210-215
Proceedings
Proceedings of ASRU 2013
Conference
IEEE 2013 Workshop on Automatic Speech Recognition and Understanding, Olomouc, CZ
ISBN
978-1-4799-2755-5
Publisher
IEEE Signal Processing Society
Place
Olomouc, CZ
DOI
10.1109/ASRU.2013.6707731
EID Scopus
2-s2.0-84893692703
BibTeX 
@INPROCEEDINGS{FITPUB10506,
   author = "Damianos Karakos and Richard Schwartz and Stavros Tsakalidis and Le Zhang and Shivesh Ranjan and Tim Ng and Roger Hsiao and Long Nguyen and Franti\v{s}ek Gr\'{e}zl and Mirko Hannemann and Martin Karafi\'{a}t and Igor Sz\H{o}ke and Karel Vesel\'{y} and et al.",
   title = "Score Normalization and System Combination for Improved Keyword Spotting",
   pages = "210--215",
   booktitle = "Proceedings of ASRU 2013",
   year = 2013,
   location = "Olomouc, CZ",
   publisher = "IEEE Signal Processing Society",
   ISBN = "978-1-4799-2755-5",
   doi = "10.1109/ASRU.2013.6707731",
   language = "english",
   url = "https://www.fit.vut.cz/research/publication/10506"
}
Back to top