The KTH synthesis of singing

This is an overview of the work with synthesizing singing that has been carried out at the Speech Music Hearing Department, KTH since 1977. The origin of the work, a hardware synthesis machine, is described and some aspects of the control program, a modified version of a text-to-speech conversion system are reviewed. Three applications are described in which the synthesis system has paved the way for investigations of specific aspects of the singing voice. One concerns the perceptual relevance of the center frequency of the singer's formant, one deals with characteristics of an ugly voice, and one regards intonation. The article is accompanied by 18 sound examples, several of which were not published before. Finally, limitations and advantages of singing synthesis are discussed.