On Inter-Signal Transplantation of Voice Characteristics
Werner Verhelst, Tim Ceyssens, Patrick Wambacq · Lirias · 2002
In this paper we study the concept of inter-signal transplantation of voice characteristics. We describe several example transplantation systems that were used to interchange selected features of the human voice amongst different utterances of a same text by a same or by different speakers. All systems are based on time-domain overlap-add (OLA) algorithms that allow high-quality prosodic modifications of speech and they all use dynamic time warping (DTW) for proper time alignment of the transplanted features. Based on these example implementations, the capabilities and the limitations of the transplantation concept are explored. Examples of successfully transplanted features are the timing and prosody of spoken and sung human voices. Possible application areas for such transplantation systems include postsynchronization in film and video industries and copying of an artist’s singing performance (‘super-karaoke’).