Affective Intonation-Modeling for Mandarin Based on PCA

Computational Linguistics and Chinese Language Processing Vol. 12, No. 1, March 2007, pp. 33-48 33 © The Association for Computational Linguistics and Chinese Language Processing [Received September 16, 2006; Revised October 10, 2006; Accepted November 1, 2006] Affective Intonation-Modeling for Mandarin Based on PCA Zhuangluan Su∗, and Zengfu Wang∗ Abstract The speech fundamental frequency (henceforth F0) contour plays an important role in expressing the affective information of an utterance. The most popular F0 modeling approaches mainly use the concept of separating the F0 contour into a global trend and local variation. For Mandarin, the global trend of the F0 contour is caused by the speaker’s mood and emotion. In this paper, the authors address the problem of affective intonation. For modeling affective intonation, an affective corpus has been designed and established, and all intonations are extracted with an iterative algorithm. Then, the concept of eigen-intonation is proposed based on the technique of Principal Component Analysis on the affective corpus and all the intonations are transformed to the lower-dimensional eigen sub-space spanned by eigen-intonations. A model of affective intonations is established in the sub-space. As a result, the corresponding emotion (maybe a mixed emotion) can be expressed by speech whose intonation is modified according to the above model. The experiments are performed with the affective Mandarin corpus, and the experimental results show that the intonation modeling approach proposed in this paper is efficient for both intonation representation and speech synthesis. Keywords: Eigen-Intonation, Affective Speech, Mixed Emotion, F0 Contour, Speech Synthesis 1. Introduction Speech can convey not only literal meanings, but also the mood and emotion of a speaker. Some researchers have proven that the contour of the speech fundamental frequency (henceforth F0 contour) plays an important role in expressing the affective information of an utterance. It is concluded that some statistical characteristics of F0 play the most important roles in emotion perception [Tao and Kang 2005]. Especially, F0 contours differ from each ∗ Department of Automation, University of Science and Technology of China, Hefei 230027, China E-mail: zfwang@ustc.edu.cn The author for correspondence is Zengfu Wang. 34 Zhuangluan Su and Zengfu Wang other because of the speaker’s different emotion in Mandarin [Yuan et al. 2002]. Due to significance of F0, the F0 contour modeling is one of the key issues that should be addressed. The most popular F0 modeling approaches mainly use the concept of separating the F0 contour into a global trend and local variation [Abe and Sato 1992; Bellegarda et al, 2001]. Mandarin is a tonal language including four basic tone types and a so-called ‘light’ tone. The F0 contour is composed of three elements [Zhao 1980]: the tone of the syllable, the variety of tone in continuous utterance, and the movement influenced by mood. How to extract tones and intonations from speech is a difficult problem. Tian and Nurminen have proposed a data-driven tone modeling approach to describe the tonal element [Tian and Nurminen 2004]. In previous work [Su and Wang 2005], the authors of this paper also proposed …

Content truncated to 500 words