Segment selection method based on tonal validity evaluation using machine learning for concatenattve speech synthesis

Akihiro Yoshida, Hideyuki Mizuno, Kazunori Mano

Research output: Chapter in Book/Report/Conference proceedingConference contribution

4 Citations (Scopus)

Abstract

This paper proposes a speech segment selection method based on machine learning for concatenative speech synthesis systems. The proposed method has two novel features. One is its use of Support Vector Machine (SVM) to estimate the subjective correctness of pitch accent with respect to each accent phrase of possible candidate speech segments. The other is its use of a determination function to identify the best segment based on SVM output. The determination function involves two assessments; one counts the number of each sign of SVM output and the other compares the distance values. The sign of SVM output is generally used to classify target objects, but the distance SVM output also represents important information. An experiment that assesses SVM performance for Japanese accent validity shows that its accuracy is 81%. To confirm the effectiveness of the proposed segment selection method, preference tests are conducted. The test indicates that the proposed method can yield Japanese synthesized speech with more natural intonation than the conventional method that uses only target cost and concatenation cost.

Original languageEnglish
Title of host publicationICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
Pages4617-4620
Number of pages4
DOIs
Publication statusPublished - 2008
Externally publishedYes
Event2008 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP - Las Vegas, NV
Duration: 2008 Mar 312008 Apr 4

Other

Other2008 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP
CityLas Vegas, NV
Period08/3/3108/4/4

Fingerprint

Speech synthesis
machine learning
Support vector machines
Learning systems
evaluation
synthesis
output
costs
Costs
estimates
Experiments

Keywords

  • Accent
  • Eoncatenative speech synthesis
  • Machine learning
  • Segment selection

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Signal Processing
  • Acoustics and Ultrasonics

Cite this

Yoshida, A., Mizuno, H., & Mano, K. (2008). Segment selection method based on tonal validity evaluation using machine learning for concatenattve speech synthesis. In ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings (pp. 4617-4620). [4518685] https://doi.org/10.1109/ICASSP.2008.4518685

Segment selection method based on tonal validity evaluation using machine learning for concatenattve speech synthesis. / Yoshida, Akihiro; Mizuno, Hideyuki; Mano, Kazunori.

ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings. 2008. p. 4617-4620 4518685.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Yoshida, A, Mizuno, H & Mano, K 2008, Segment selection method based on tonal validity evaluation using machine learning for concatenattve speech synthesis. in ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings., 4518685, pp. 4617-4620, 2008 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP, Las Vegas, NV, 08/3/31. https://doi.org/10.1109/ICASSP.2008.4518685
Yoshida A, Mizuno H, Mano K. Segment selection method based on tonal validity evaluation using machine learning for concatenattve speech synthesis. In ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings. 2008. p. 4617-4620. 4518685 https://doi.org/10.1109/ICASSP.2008.4518685
Yoshida, Akihiro ; Mizuno, Hideyuki ; Mano, Kazunori. / Segment selection method based on tonal validity evaluation using machine learning for concatenattve speech synthesis. ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings. 2008. pp. 4617-4620
@inproceedings{a80ad12d36c7414ead4808770df4d11d,
title = "Segment selection method based on tonal validity evaluation using machine learning for concatenattve speech synthesis",
abstract = "This paper proposes a speech segment selection method based on machine learning for concatenative speech synthesis systems. The proposed method has two novel features. One is its use of Support Vector Machine (SVM) to estimate the subjective correctness of pitch accent with respect to each accent phrase of possible candidate speech segments. The other is its use of a determination function to identify the best segment based on SVM output. The determination function involves two assessments; one counts the number of each sign of SVM output and the other compares the distance values. The sign of SVM output is generally used to classify target objects, but the distance SVM output also represents important information. An experiment that assesses SVM performance for Japanese accent validity shows that its accuracy is 81{\%}. To confirm the effectiveness of the proposed segment selection method, preference tests are conducted. The test indicates that the proposed method can yield Japanese synthesized speech with more natural intonation than the conventional method that uses only target cost and concatenation cost.",
keywords = "Accent, Eoncatenative speech synthesis, Machine learning, Segment selection",
author = "Akihiro Yoshida and Hideyuki Mizuno and Kazunori Mano",
year = "2008",
doi = "10.1109/ICASSP.2008.4518685",
language = "English",
isbn = "1424414849",
pages = "4617--4620",
booktitle = "ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings",

}

TY - GEN

T1 - Segment selection method based on tonal validity evaluation using machine learning for concatenattve speech synthesis

AU - Yoshida, Akihiro

AU - Mizuno, Hideyuki

AU - Mano, Kazunori

PY - 2008

Y1 - 2008

N2 - This paper proposes a speech segment selection method based on machine learning for concatenative speech synthesis systems. The proposed method has two novel features. One is its use of Support Vector Machine (SVM) to estimate the subjective correctness of pitch accent with respect to each accent phrase of possible candidate speech segments. The other is its use of a determination function to identify the best segment based on SVM output. The determination function involves two assessments; one counts the number of each sign of SVM output and the other compares the distance values. The sign of SVM output is generally used to classify target objects, but the distance SVM output also represents important information. An experiment that assesses SVM performance for Japanese accent validity shows that its accuracy is 81%. To confirm the effectiveness of the proposed segment selection method, preference tests are conducted. The test indicates that the proposed method can yield Japanese synthesized speech with more natural intonation than the conventional method that uses only target cost and concatenation cost.

AB - This paper proposes a speech segment selection method based on machine learning for concatenative speech synthesis systems. The proposed method has two novel features. One is its use of Support Vector Machine (SVM) to estimate the subjective correctness of pitch accent with respect to each accent phrase of possible candidate speech segments. The other is its use of a determination function to identify the best segment based on SVM output. The determination function involves two assessments; one counts the number of each sign of SVM output and the other compares the distance values. The sign of SVM output is generally used to classify target objects, but the distance SVM output also represents important information. An experiment that assesses SVM performance for Japanese accent validity shows that its accuracy is 81%. To confirm the effectiveness of the proposed segment selection method, preference tests are conducted. The test indicates that the proposed method can yield Japanese synthesized speech with more natural intonation than the conventional method that uses only target cost and concatenation cost.

KW - Accent

KW - Eoncatenative speech synthesis

KW - Machine learning

KW - Segment selection

UR - http://www.scopus.com/inward/record.url?scp=51549102206&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=51549102206&partnerID=8YFLogxK

U2 - 10.1109/ICASSP.2008.4518685

DO - 10.1109/ICASSP.2008.4518685

M3 - Conference contribution

AN - SCOPUS:51549102206

SN - 1424414849

SN - 9781424414840

SP - 4617

EP - 4620

BT - ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings

ER -