Second ESCA/IEEE Workshop on Speech Synthesis
September 12-15, 1994
When generating synthetic speech by unit concatenation a major point is the definition of the unit inventory. Traditionally, diphone or demisyllable inventories are used but both unit types have their drawbacks. Based on the results of a comprehensive investigation of coarticulatory phenomena at syllable boundaries, a new mixed inventory structure was developed which is syllable oriented but does not demand a definite decision about the position of a syllable boundary. A preliminary evaluation comparing the mixed with a demisyllable and a diphone inventory confirms that indeed some of the disadvantages of the two standard methods vanish when the mixed inventory structure is used. A complete inventory was established that consists of approximately 2200 units for the German language.
Bibliographic reference. Portele, Thomas / Höfer, Florian / Hess, Wolfgang (1994): "A mixed inventory structure for German concatenative synthesis", In SSW2-1994, 115-118.