Publication detail

Speech units automatically generated by ergodic hidden Markov model

SZŐKE, I.

Original Title

Speech units automatically generated by ergodic hidden Markov model

English Title

Speech units automatically generated by ergodic hidden Markov model

Type

conference paper

Language

en

Original Abstract

Units automatically generated are used in VLBR (Very Low BitRate) coding. One approach is to generate ALISP - Automatic Language Independent Speech Processing units by temporal decomposition and vector quantization were used there. Our goal is to propose better approach. We use EHMM - Ergodic Hidden Markov Model for ALISP units determination. A brief description of the model, its initialization and training is discussed in this paper. Experimental results are discussed in the conclusion.

English abstract

Units automatically generated are used in VLBR (Very Low BitRate) coding. One approach is to generate ALISP - Automatic Language Independent Speech Processing units by temporal decomposition and vector quantization were used there. Our goal is to propose better approach. We use EHMM - Ergodic Hidden Markov Model for ALISP units determination. A brief description of the model, its initialization and training is discussed in this paper. Experimental results are discussed in the conclusion.

Keywords

speech, units, automatically, generated, ergodic, hidden Markov model,

Released

18.04.2004

Publisher

Faculty of Electrical Engineering and Communication BUT

Location

Brno

Pages from

1

Pages to

5

Pages count

5

URL

BibTex


@inproceedings{BUT16913,
  author="Igor {Szőke}",
  title="Speech units automatically generated by ergodic hidden Markov model",
  annote="Units automatically generated are used in VLBR (Very Low BitRate)
coding. One approach is to generate ALISP - Automatic Language
Independent Speech Processing units by temporal decomposition and
vector quantization were used there. Our goal is to propose better
approach. We use EHMM - Ergodic Hidden Markov Model for ALISP units
determination. A brief description of the model, its initialization and
training is discussed in this paper. Experimental results are discussed
in the conclusion.",
  address="Faculty of Electrical Engineering and Communication BUT",
  booktitle="Proceedings of 10th Conference and Competition STUDENT EEICT 2004",
  chapter="16913",
  institution="Faculty of Electrical Engineering and Communication BUT",
  year="2004",
  month="april",
  pages="1",
  publisher="Faculty of Electrical Engineering and Communication BUT",
  type="conference paper"
}