Generation and perception of F0 markedness for communicative speech synthesis

Aiming at natural F0 control for conversational speech synthesis using attributes of constituent output words, F0 characteristics are analyzed from both generation and perception viewpoints. We recorded commonly used two-phrase utterances consisting of Japanese adjective and adverb phrases expressing different degree of markedness under designed conversational situations, and compared their F0 characteristics. The comparison showed the consistent F0 control dependencies not only on adverbs themselves but also on the attribute of following adjective phrases. Strong positive or negative correlation is observed between the markedness of adverbs and F0 height when an adjective phrase showing positiveness or negativeness is followed to the current adverb phrase. These consistencies have been perceptually confirmed by naturalness evaluation tests using the same two-phrase samples with different F0 heights. Finally, a computational model of conversational F0 control is proposed using lexical information of adjectives showing positiveness or negativeness and adverbs expressing markedness. F0 estimation experiments quantitatively showed the possibility of F0 control for natural conversational speech synthesis using the attribute of constituent output words.