Tree-Based Approaches to Automatic Generation of Speech Synthesis Rules for Prosodic Parameters

Yoichi YAMASHITA; Manabu TANAKA; Yoshitake AMAKO; Yasuo NOMURA; Yoshikazu OHTA; Atsunori KITOH; Osamu KAKUSHO; Riichiro MIZOGUCHI

Tree-Based Approaches to Automatic Generation of Speech Synthesis Rules for Prosodic Parameters

Yoichi YAMASHITA, Manabu TANAKA, Yoshitake AMAKO, Yasuo NOMURA, Yoshikazu OHTA, Atsunori KITOH, Osamu KAKUSHO, Riichiro MIZOGUCHI

Full Text Views

0

Share
Cite this

Summary :

This paper describes automatic generation of speech synthesis rules which predict a stress level for each bunsetsu in long noun phrases. The rules are inductively inferred from a lot of speech data by using two kinds of tree-based methods, the conventional decision tree and the SBR-tree methods. The rule sets automatically generated by two methods have almost the same performance and decrease the prediction error to about 14 Hz from 23 Hz of the accent component value. The rate of the correct reproduction of the change for adjacent bunsetsu pairs is also used as a measure for evaluating the generated rule sets and they correctly reproduce the change of about 80%. The effectiveness of the rule sets is verified through the listening test. And, with regard to the comprehensiveness of the generated rules, the rules by the SBR-tree methods are very compact and easy to human experts to interpret and matches the former studies.

Publication: IEICE TRANSACTIONS on Fundamentals Vol.E76-A No.11 pp.1934-1941

Publication Date: 1993/11/25

Publicized

Online ISSN

DOI

Type of Manuscript: Special Section PAPER (Special Section on Speech Synthesis: Current Technologies and Thier Application)

Category

Cite this

Copy

Yoichi YAMASHITA, Manabu TANAKA, Yoshitake AMAKO, Yasuo NOMURA, Yoshikazu OHTA, Atsunori KITOH, Osamu KAKUSHO, Riichiro MIZOGUCHI, "Tree-Based Approaches to Automatic Generation of Speech Synthesis Rules for Prosodic Parameters" in IEICE TRANSACTIONS on Fundamentals, vol. E76-A, no. 11, pp. 1934-1941, November 1993, doi: .
Abstract: This paper describes automatic generation of speech synthesis rules which predict a stress level for each bunsetsu in long noun phrases. The rules are inductively inferred from a lot of speech data by using two kinds of tree-based methods, the conventional decision tree and the SBR-tree methods. The rule sets automatically generated by two methods have almost the same performance and decrease the prediction error to about 14 Hz from 23 Hz of the accent component value. The rate of the correct reproduction of the change for adjacent bunsetsu pairs is also used as a measure for evaluating the generated rule sets and they correctly reproduce the change of about 80%. The effectiveness of the rule sets is verified through the listening test. And, with regard to the comprehensiveness of the generated rules, the rules by the SBR-tree methods are very compact and easy to human experts to interpret and matches the former studies.
URL: https://globals.ieice.org/en_transactions/fundamentals/10.1587/e76-a_11_1934/_p

Copy

@ARTICLE{e76-a_11_1934,
author={Yoichi YAMASHITA, Manabu TANAKA, Yoshitake AMAKO, Yasuo NOMURA, Yoshikazu OHTA, Atsunori KITOH, Osamu KAKUSHO, Riichiro MIZOGUCHI, },
journal={IEICE TRANSACTIONS on Fundamentals},
title={Tree-Based Approaches to Automatic Generation of Speech Synthesis Rules for Prosodic Parameters},
year={1993},
volume={E76-A},
number={11},
pages={1934-1941},
abstract={This paper describes automatic generation of speech synthesis rules which predict a stress level for each bunsetsu in long noun phrases. The rules are inductively inferred from a lot of speech data by using two kinds of tree-based methods, the conventional decision tree and the SBR-tree methods. The rule sets automatically generated by two methods have almost the same performance and decrease the prediction error to about 14 Hz from 23 Hz of the accent component value. The rate of the correct reproduction of the change for adjacent bunsetsu pairs is also used as a measure for evaluating the generated rule sets and they correctly reproduce the change of about 80%. The effectiveness of the rule sets is verified through the listening test. And, with regard to the comprehensiveness of the generated rules, the rules by the SBR-tree methods are very compact and easy to human experts to interpret and matches the former studies.},
keywords={},
doi={},
ISSN={},
month={November},}

Copy

TY - JOUR
TI - Tree-Based Approaches to Automatic Generation of Speech Synthesis Rules for Prosodic Parameters
T2 - IEICE TRANSACTIONS on Fundamentals
SP - 1934
EP - 1941
AU - Yoichi YAMASHITA
AU - Manabu TANAKA
AU - Yoshitake AMAKO
AU - Yasuo NOMURA
AU - Yoshikazu OHTA
AU - Atsunori KITOH
AU - Osamu KAKUSHO
AU - Riichiro MIZOGUCHI
PY - 1993
DO -
JO - IEICE TRANSACTIONS on Fundamentals
SN -
VL - E76-A
IS - 11
JA - IEICE TRANSACTIONS on Fundamentals
Y1 - November 1993
AB - This paper describes automatic generation of speech synthesis rules which predict a stress level for each bunsetsu in long noun phrases. The rules are inductively inferred from a lot of speech data by using two kinds of tree-based methods, the conventional decision tree and the SBR-tree methods. The rule sets automatically generated by two methods have almost the same performance and decrease the prediction error to about 14 Hz from 23 Hz of the accent component value. The rate of the correct reproduction of the change for adjacent bunsetsu pairs is also used as a measure for evaluating the generated rule sets and they correctly reproduce the change of about 80%. The effectiveness of the rule sets is verified through the listening test. And, with regard to the comprehensiveness of the generated rules, the rules by the SBR-tree methods are very compact and easy to human experts to interpret and matches the former studies.
ER -