Browse > Article

Generating Korean Energy Contours Using Vector-regression Tree  

이상호 (LG전자기술원 모바일 멀티미디어 연구소)
오영환 (한국과학기술원 전자전산학과 전산학전공)
Abstract
This study describes an energy contour generation method for Korean n systems. We propose a vector-regression tree, which is a vector version of a scalar regression tree. A vector-regression tree predicts a response vector for an unknown feature vector. In our study, the tree yields a vector containing ten sampled energy values for each phone. After collecting 500 sentences and its corresponding speech corpus, we trained trees on 300 sentences and tested them on 200 sentences. We construct a bagged tree and a born again one to improve the performance of contour prediction. In the experiment, we got a 0.803 correlation coefficient for the observed and predicted energy values.
Keywords
Energy contour; TTS system; Vector-regression tree; Prosody; Speech synthesis;
Citations & Related Records
연도 인용수 순위
  • Reference
1 Bagging Predictors /
[ L.Breiman ] / Machine Learning
2 /
[ L.Breiman;J.H.Friedman;R.A.Olshen;C.J.Stone ] / Classification and Regression Trees ser. Wadsworth Statistics/Probability Series
3 Unsupervised training of phone duration and energy models for text-to-speech synthesis /
[ P.C.Bagshaw ] / Porc. Int. Conf. Spoken Language Processing
4 Optimal partitioning for classification and regression trees /
[ P.A.Chou ] / IEEE Trans. Pattern Anal. Machine Intell.   DOI   ScienceOn
5 /
[ L.Breiman ] / Out-of-Bag Estimation
6 Energy contour generation for a sentence using a neural network learning method /
[ J.C.Lee;D.G.Kang;S.H.Kim;K.M.Sung ] / Proc. Int. Conf. Spoken Language Processing
7 A dynamical system model for generating fundamental frequency for speech synthesis /
[ K.Ross;M.Ostendorf ] / IEEE Trans. Speech Audio Processing   DOI   ScienceOn
8 /
[ L.Breiman;N.Shang ] / Born Again Trees
9 한국어 억양의 트리 기반 모델링 /
[ 이상호;오영환 ] / 한국음향학회지