Speech Prosody 2008
This paper describes an investigation into the repair of the prosodic limitations of tracheoesophageal (TE) speech. The proposed repair algorithm modifies TE phone durations based on the predictions of regression trees built from nonpathological data. Acoustic and language modelling refinements for improved TE phone recognition, studies of feature relevance for duration prediction and a robust duration modification method are also presented. Objective and subjective evaluation of results show that the duration pattern of the repaired sentences is closer to normal and perceptually preferred to the original in terms of overall rhythmic naturalness.
Bibliographic reference. Pozo, Arantza del / Young, Steve (2008): "Repairing tracheoesophageal speech duration", In SP-2008, 187-190.