14th International Congress of Phonetic Sciences (ICPhS-14)
San Francisco, CA, USA
Our Hebrew speech synthesis (although based entirely on rules)
attempts to resemble natural speech production. Our speech
communication model assumes that speech (both in production
and perception) is constructed from sequential segments, which
form a continuum. Communication occurs simultaneously on
various levels, which can be differentiated by the length of their
segments. The basic segment length is that of the phoneme.
Higher levels, consisting of longer segments (syllables, stress units,
sense-groups, breath-groups, and intonation-groups),
conform to associated rules that affect the lower levels.
The F0 contour is chosen to demonstrate synthesis rules derived from our model. It is shown how F0 can be created step by step, taking into account the different layers of speech production or perception. We demonstrate how F0 is molded as a result of the influence of a few levels and how mutual effects among levels lead to the construction of the composite fundamental frequency contour.
Bibliographic reference. Laufer, Asher / Sonnenblick, Yehuda / Teitelbaum, Aryeh (1999): "Fundamental frequency contour in Hebrew synthesis by rule an imitation of a speech communication model", In ICPhS-14, 65-68.