Publication Details

Phoneme Based Acoustics Keyword Spotting in Informal Continuous Speech

SZŐKE Igor, SCHWARZ Petr, BURGET Lukáš, KARAFIÁT Martin, MATĚJKA Pavel and ČERNOCKÝ Jan. Phoneme Based Acoustics Keyword Spotting in Informal Continuous Speech. Lecture Notes in Computer Science, vol. 2005, no. 3658, p. 8. ISSN 0302-9743.
Czech title
Fonémový detektor klíčových slov založený na akustice pro neformální konverzační řeč
Type
journal article
Language
english
Authors
Szőke Igor, Ing., Ph.D. (DCGM FIT BUT)
Schwarz Petr, Ing., Ph.D. (DCGM FIT BUT)
Burget Lukáš, doc. Ing., Ph.D. (DCGM FIT BUT)
Karafiát Martin, Ing., Ph.D. (DCGM FIT BUT)
Matějka Pavel, Ing., Ph.D. (DCGM FIT BUT)
Černocký Jan, prof. Dr. Ing. (DCGM FIT BUT)
URL
Keywords

acoustic keyword spotting, hidden Markov model, phoneme, recognition network

Abstract

This paper describes several ways of acoustic keywords spotting (KWS), based on Gaussian mixture model (GMM) hidden Markov models (HMM) and phoneme posterior probabilities from FeatureNet. Context-independent and dependent phoneme models are used in the GMM/HMM system. The systems were trained and evaluated on informal continuous speech. We used different complexities of KWS recognition network and different types of phoneme models. We study the impact of these parameters on the accuracy and computational complexity, and conclude that phoneme posteriors outperform conventional GMM/HMM system.

Published
2005
Pages
8
Journal
Lecture Notes in Computer Science, vol. 2005, no. 3658, ISSN 0302-9743
Publisher
Springer Verlag
BibTeX 
@ARTICLE{FITPUB7882,
   author = "Igor Sz\H{o}ke and Petr Schwarz and Luk\'{a}\v{s} Burget and Martin Karafi\'{a}t and Pavel Mat\v{e}jka and Jan \v{C}ernock\'{y}",
   title = "Phoneme Based Acoustics Keyword Spotting in Informal Continuous Speech",
   pages = 8,
   journal = "Lecture Notes in Computer Science",
   volume = 2005,
   number = 3658,
   year = 2005,
   ISSN = "0302-9743",
   language = "english",
   url = "https://www.fit.vut.cz/research/publication/7882"
}
Back to top