Prosody
Text-To-Speech synthesis

Prosody is the "music" of speech, i.e. its melodic, rhythmic and dynamic components. The same string of phonemes can be pronounced in very different "tones", thanks to prosodic variation. From a synthesis point of view, prosodic computation consists in modeling and predicting :

• melodic contours, through the temporal evolution of the fundamental frequency of vocal cord vibration;

• syllabic rhythm, through syllable and phoneme durations;

• the rhythm of word groups, through the position and duration of pauses;

• dynamics, vocal effort, relative syllable intensity;

• vocal quality, including murmuring, whispering and different voicing mechanisms;

• extralinguistic elements, such as sighs,...