Publication detail

Prosodic Modification of Synthetic Speech

SZŐKE, I.

Original Title

Prosodic Modification of Synthetic Speech

English Title

Prosodic Modification of Synthetic Speech

Type

conference paper

Language

en

Original Abstract

This article presents a method for speech prosodic modification. Parametric harmonic and noise model (HNM) is used to describe a speech signal. The algorithm for prosodic modifications consists of two parts. The analysis part parametrizes the speech signal. We spread speech into a sequence of the pitch-marks, each of which is presented by a vector of parameters. The next part is synthesis. Prosodic modifications are done by remapping of the pitch-marks. Modified signal can be resynthetized from interpolated parameters.

English abstract

This article presents a method for speech prosodic modification. Parametric harmonic and noise model (HNM) is used to describe a speech signal. The algorithm for prosodic modifications consists of two parts. The analysis part parametrizes the speech signal. We spread speech into a sequence of the pitch-marks, each of which is presented by a vector of parameters. The next part is synthesis. Prosodic modifications are done by remapping of the pitch-marks. Modified signal can be resynthetized from interpolated parameters.

Keywords

prosody, modification, synthetic, speech, harmonic, noise, model, pitch

RIV year

2003

Released

26.04.2003

Publisher

Faculty of Electrical Engineering and Communication BUT

Location

Brno

ISBN

80-214-2401-X

Book

Proceedings of The International Conference and Competition STUDENT EEICT 2003

Pages from

349

Pages to

352

Pages count

4

URL

Documents

BibTex


@inproceedings{BUT10912,
  author="Igor {Szőke}",
  title="Prosodic Modification of Synthetic Speech",
  annote="This article presents a method for speech prosodic modification.
Parametric harmonic and noise model (HNM) is used to describe a speech
signal. The algorithm for prosodic modifications consists of two parts.
The analysis part parametrizes the speech signal. We spread speech into
a sequence of the pitch-marks, each of which is presented by a vector
of parameters. The next part is synthesis. Prosodic modifications are
done by remapping of the pitch-marks. Modified signal can be
resynthetized from interpolated parameters.
", address="Faculty of Electrical Engineering and Communication BUT", booktitle="Proceedings of The International Conference and Competition STUDENT EEICT 2003", chapter="10912", institution="Faculty of Electrical Engineering and Communication BUT", year="2003", month="april", pages="349--352", publisher="Faculty of Electrical Engineering and Communication BUT", type="conference paper" }