Detail publikace

Phoneme Based Acoustics Keyword Spotting in Informal Continuous Speech

SZŐKE, I., SCHWARZ, P., BURGET, L., KARAFIÁT, M., MATĚJKA, P., ČERNOCKÝ, J.

Originální název

Phoneme Based Acoustics Keyword Spotting in Informal Continuous Speech

Anglický název

Phoneme Based Acoustics Keyword Spotting in Informal Continuous Speech

Jazyk

en

Originální abstrakt

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.

Anglický abstrakt

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.

Dokumenty

BibTex


@article{BUT42913,
  author="Igor {Szőke} and Petr {Schwarz} and Lukáš {Burget} and Martin {Karafiát} and Pavel {Matějka} and Jan {Černocký}",
  title="Phoneme Based Acoustics Keyword Spotting in Informal Continuous Speech",
  annote="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.",
  chapter="42913",
  journal="Lecture Notes in Computer Science (IF 0,513)",
  number="3658",
  volume="2005",
  year="2005",
  month="august",
  pages="302",
  type="journal article - other"
}