Voice Conversion: State of the art and Perspectives
by Yannis Styilianou (University of Crete, Grece)

Voice Transformation refers to the various modifications one may apply to the sound produced by a person, speaking or singing. In other words, Voice Transformation aims at the control of non-linguistic information of speech signals such as voice quality and voice individuality. Voice Transformation covers a wide area of research from speech production modeling and understanding to perception of speech, from natural language processing, modeling and control of speaking style, to pattern recognition and statistical signal processing.

Voice Transformation was considered as a hot, novel and fast-growing topic in 1990s having as potential application the concatenated speech synthesis systems where new (virtual or target) voices could be created without requiring to pass through the quite expensive process of developing new voices. By that time, it was widely accepted that Voice Transformation systems were far from providing the required performance. With the recent developments in speech synthesis this need is more pronounced. There is an increasing demand for high quality Voice Transformation methods not only for creating target or virtual voices, but also to model various effects (e.g., Lombard effect), synthesize emotions, to make more natural the dialog systems which use speech synthesis etc.

In this talk I will review the state-of-the-art Voice Transformation methodology showing its limitations in producing good speech quality and its current challenges. Addressing quality issues of current voice transformation algorithms in conjunction with properties of the speech production and speech perception systems I will try to pave the way for more natural Voice Transformation algorithms in the future. Facing the challenges, it will allow Voice Transformation systems to be applied in important and versatile areas of speech technology. Besides speech synthesis, Voice Transformation has other potential applications in areas like entertainment, film, and music industry, toys, chat rooms and games, dialog systems, security and speaker individuality for interpreting telephony, high-end hearing aids, vocal pathology and voice restoration.

Yannis Stylianou is Associate Professor at University of Crete, Department of Computer Science. He received the Diploma of Electrical Engineering from the National Technical University, NTUA, of Athens in 1991 and the M.Sc. and Ph.D. degrees in Signal Processing from the Ecole National Superieure des Telecommunications, ENST, Paris, France in 1992 and 1996, respectively. From 1996 until 2001 he was with AT&T Labs Research (Murray Hill and Florham Park, NJ, USA) as a Senior Technical Staff Member. In 2001 he joined Bell-Labs Lucent Technologies, in Murray Hill, NJ, USA.

Since 2002 he is with the Computer Science Department at the University of Crete. He was Associate Editor for the IEEE Signal Processing Letters from 1999 until 2002. He is Associate Editor of the EURASIP Journal on Speech, Audio and Music Processing. He was served on the Management Committee for the COST Action 277: "Nonlinear Speech Processing" and he is one of the two proponents for a new COST Action on Voice Quality Assessment. He holds 8 patents. Prof. Stylianou participates in the SIMILAR Network of Excellence (6th FP) coordinating the task on the fusion of speech and handwriting modalities.

Invited Conference 1