Boundary detection using continuous wavelet analysis

Antti Suni, Juraj Šimko, Martti Vainio

Research output: Chapter in Book/Report/Conference proceedingConference contributionScientificpeer-review

Abstract

Unsupervised boundary detection and classification is both a theoretically interesting question and an important challenge for speech technology. Theoretical interest lies in exploring how and to what extent is the boundary information encoded in purely acoustic material. For technology, automatic boundary detection facilitates cheap and fast labeling of large corpora of speech data. In this work we present a novel methodology of automatic and unsupervised boundary detection and classification based on the continuous wavelet transform (CWT) technique. Several approaches using lines of minimal amplitude, phase information and wavelet-based estimation of speech tempo are evaluated and compared on Boston Radio News Corpus data. The results show that this methodology using hierarchical information encoded in speech signal compares favorably with traditionally used supervised boundary detection techniques using acoustic information.

Original languageEnglish
Title of host publicationProceedings of the International Conference on Speech Prosody
PublisherInternational Speech Communication Association
Pages267-271
Number of pages5
Volume2016-January
DOIs
Publication statusPublished - 2016
MoE publication typeA4 Article in a conference publication
EventInternational Conference on Speech Prosody - Boston, United States
Duration: 31 May 20163 Jun 2016
Conference number: 8

Publication series

NameSpeech prosody : an international conference
PublisherInternational Speech Communication Association
ISSN (Electronic)2333-2042

Conference

ConferenceInternational Conference on Speech Prosody
CountryUnited States
CityBoston
Period31/05/201603/06/2016

Keywords

  • Boundary detection
  • Continuous wavelet transform
  • Speech synthesis

Fingerprint Dive into the research topics of 'Boundary detection using continuous wavelet analysis'. Together they form a unique fingerprint.

Cite this