Detail publikačního výsledku

Sound-Dependent Spectral Warping in Voice Identity Conversion

Martin Vondra, Robert Vích

Originální název

Sound-Dependent Spectral Warping in Voice Identity Conversion

Anglický název

Sound-Dependent Spectral Warping in Voice Identity Conversion

Druh

Stať ve sborníku v databázi WoS či Scopus

Originální abstrakt

In this paper a new technique for voice conversion is introduced, where different spectral warping functions for different frames of the source speech are used. Warping functions are obtained by averaging the optimum frequency transformation function between the appropriate source and target short-time smoothed spectra of time-aligned training speech sequences.

Anglický abstrakt

In this paper a new technique for voice conversion is introduced, where different spectral warping functions for different frames of the source speech are used. Warping functions are obtained by averaging the optimum frequency transformation function between the appropriate source and target short-time smoothed spectra of time-aligned training speech sequences.

Klíčová slova

Dynamic spectral warping, speech analysis, speech synthesis, cepstrum, vector quantization

Klíčová slova v angličtině

Dynamic spectral warping, speech analysis, speech synthesis, cepstrum, vector quantization

Autoři

Martin Vondra, Robert Vích

Vydáno

01.01.2005

Nakladatel

TUD press

Místo

Dresden

ISBN

3-938863-17-X

Kniha

Electronic Speech Signal Processing

Strany od

423

Strany počet

7

BibTex

@inproceedings{BUT15226,
  author="Martin {Vondra} and Robert {Vích}",
  title="Sound-Dependent Spectral Warping in Voice Identity Conversion",
  booktitle="Electronic Speech Signal Processing",
  year="2005",
  pages="7",
  publisher="TUD press",
  address="Dresden",
  isbn="3-938863-17-X"
}