Techniques for Spectral Voice Conversion
In his thesis, Victor Popa focuses on improving the voice conversion methods in terms of both speech quality and identity mapping. Voice conversion is a relatively new topic in speech research aiming essentially to change the perceived voice in speech signals.
In the context of booming modern technology, the presence of computers in our lives became a fact and brought the need for a more natural communication between humans and machines. Speech is the most accurate and natural communication instrument between humans enabling them not only to exchange ideas but also to transmit emotions. As such, speech has been receiving a huge interest from the scientific community and many research areas have emerged to study its structure, production and perception in an effort to design speech based human computer interfaces.
Voice conversion is a relatively new topic in speech research aiming essentially to change the perceived voice in speech signals. Voice conversion technology opens the way to a large number of applications most of which are closely related to speech synthesis. The applications range from personalized text to speech voices to entertainment and security related applications. Potential applications in the entertainment industry include movie dubbing with the original artist voices, generating speech with voices that no longer exist, disguising speaker identity, creating virtual voices for characters in a videogame or creating compact “audio books” in text format where the narrator and characters involved in dialogues each use their individual voice. Furthermore e-mails and SMS messages could be “read” to us with the sender’s voice.
In the thesis “Techniques for Spectral Voice Conversion” the goal is to develop voice conversion algorithms capable to work with coded speech in order to facilitate the use of this technology in communications and embedded applications. At the same time, the thesis is focused on improving the voice conversion methods in terms of both speech quality and identity mapping.
Public defense of a doctoral dissertation on Friday, 1 June
The doctoral dissertation of MSc Victor Popa in the field of voice conversion titled “Techniques for Spectral Voice Conversion” will be publicly examined at the Faculty of Computing and Electrical Engineering of Tampere University of Technology (TUT) in room TB104 in Tietotalo building (Korkeakoulunkatu 1, FI-33720 Tampere) on 1.6.2012 at 12:00.
The opponent will be Professor Yannis Stylianou from the University of Crete, Greece. Professor Ioan Tabus form TUT’s Department of Signal Processing will act as Chairman.
Victor Popa (31) comes from Bucharest, Romania and works as a researcher in the Audio Group of the Department of Signal Processing at Tampere University of Technology.
Further information: Victor Popa, +358 45 264 1822, victor.popa@tut.fi