Hierarchical Representation of Prosody for Statistical Speech Synthesis

Antti Suni, Daniel Aalto, Martti Vainio

Research output: Contribution to journalArticleScientific

Abstract

Prominences and boundaries are the essential constituents of prosodic structure in speech. They provide for means to chunk the speech stream into linguistically relevant units by providing them with relative saliences and demarcating them within coherent utterance structures. Prominences and boundaries have both been widely used in both basic research on prosody as well as in text-to-speech synthesis. However, there are no representation schemes that would provide for both estimating and modelling them in a unified fashion. Here we present an unsupervised unified account for estimating and representing prosodic prominences and boundaries using a scale-space analysis based on continuous wavelet transform. The methods are evaluated and compared to earlier work using the Boston University Radio News corpus. The results show that the proposed method is comparable with the best published supervised annotation methods.
Original languageEnglish
JournalarXiv.org
VolumearXiv:1510.01949
Number of pages22
ISSN2331-8422
Publication statusPublished - 8 Sep 2015
MoE publication typeB1 Journal article

Fields of Science

  • 6161 Phonetics

Cite this

@article{9be12d48cd5044c98459e9816c08c4da,
title = "Hierarchical Representation of Prosody for Statistical Speech Synthesis",
abstract = "Prominences and boundaries are the essential constituents of prosodic structure in speech. They provide for means to chunk the speech stream into linguistically relevant units by providing them with relative saliences and demarcating them within coherent utterance structures. Prominences and boundaries have both been widely used in both basic research on prosody as well as in text-to-speech synthesis. However, there are no representation schemes that would provide for both estimating and modelling them in a unified fashion. Here we present an unsupervised unified account for estimating and representing prosodic prominences and boundaries using a scale-space analysis based on continuous wavelet transform. The methods are evaluated and compared to earlier work using the Boston University Radio News corpus. The results show that the proposed method is comparable with the best published supervised annotation methods.",
keywords = "6161 Phonetics",
author = "Antti Suni and Daniel Aalto and Martti Vainio",
year = "2015",
month = "9",
day = "8",
language = "English",
volume = "arXiv:1510.01949",
journal = "arXiv.org",
issn = "2331-8422",
publisher = "Cornell University",

}

Hierarchical Representation of Prosody for Statistical Speech Synthesis. / Suni, Antti; Aalto, Daniel; Vainio, Martti.

In: arXiv.org , Vol. arXiv:1510.01949, 08.09.2015.

Research output: Contribution to journalArticleScientific

TY - JOUR

T1 - Hierarchical Representation of Prosody for Statistical Speech Synthesis

AU - Suni, Antti

AU - Aalto, Daniel

AU - Vainio, Martti

PY - 2015/9/8

Y1 - 2015/9/8

N2 - Prominences and boundaries are the essential constituents of prosodic structure in speech. They provide for means to chunk the speech stream into linguistically relevant units by providing them with relative saliences and demarcating them within coherent utterance structures. Prominences and boundaries have both been widely used in both basic research on prosody as well as in text-to-speech synthesis. However, there are no representation schemes that would provide for both estimating and modelling them in a unified fashion. Here we present an unsupervised unified account for estimating and representing prosodic prominences and boundaries using a scale-space analysis based on continuous wavelet transform. The methods are evaluated and compared to earlier work using the Boston University Radio News corpus. The results show that the proposed method is comparable with the best published supervised annotation methods.

AB - Prominences and boundaries are the essential constituents of prosodic structure in speech. They provide for means to chunk the speech stream into linguistically relevant units by providing them with relative saliences and demarcating them within coherent utterance structures. Prominences and boundaries have both been widely used in both basic research on prosody as well as in text-to-speech synthesis. However, there are no representation schemes that would provide for both estimating and modelling them in a unified fashion. Here we present an unsupervised unified account for estimating and representing prosodic prominences and boundaries using a scale-space analysis based on continuous wavelet transform. The methods are evaluated and compared to earlier work using the Boston University Radio News corpus. The results show that the proposed method is comparable with the best published supervised annotation methods.

KW - 6161 Phonetics

M3 - Article

VL - arXiv:1510.01949

JO - arXiv.org

JF - arXiv.org

SN - 2331-8422

ER -