Publication detail

Speech Segmentation into Phonemes by Wavelet Transform

SYSEL, P.

Original Title

Speech Segmentation into Phonemes by Wavelet Transform

Type

conference paper

Language

English

Original Abstract

Speech signal consists of phonemes in which the signal has a quasistationary character. The boundaries of phonemes are currently detected by observing such attributes as signal energy, short-time spectrum, etc. We can use a time-spectral analysis as a method of determining the attributes, e.g. multiscale wavelet analysis. Then we can determine the boundaries of phonemes by observing the changes in the individual scales and by comparing their positions.

Keywords

Wavelet transformation, phonemes, speech processing, segmentation

Authors

SYSEL, P.

RIV year

2000

Released

6. 9. 2000

Publisher

FEI VUT Brno

Location

Brno

ISBN

80-7204-161-4

Book

Proceedings of the 23rd International Conference TELECOMMUNICATIONS AND SIGNAL PROCESSING TSP-2000

Pages from

126

Pages to

129

Pages count

4

BibTex

@inproceedings{BUT2298,
  author="Petr {Sysel}",
  title="Speech Segmentation into Phonemes by Wavelet Transform",
  booktitle="Proceedings of the 23rd International Conference TELECOMMUNICATIONS AND SIGNAL PROCESSING TSP-2000",
  year="2000",
  pages="126--129",
  publisher="FEI VUT Brno",
  address="Brno",
  isbn="80-7204-161-4"
}