Estimation and generalization of multimodal speech production

The speech acoustics and the phonetically relevant motion of the face during speech are determined by the time-varying behavior of the vocal tract. A benefit of this linkage is that face motion can be predicted from the spectral acoustics during sentence production. However, the scope of reliable estimation appears to be limited to individual sentences, because the analysis degrades sharply when multiple sentences are analyzed together, suggesting sentence length boundary constraints. These constraints are examined in this paper.