| 摘要 |
The invention provides a method and apparatus for providing real-time visual representations of a speaker's facial movements, associated with an acoustic speech signal delivered by the speaker, in which the acoustic speech signal is analyzed to identify its constituent elemental speech units, and an image of moving human facial features is generated and displayed on a display screen. The facial feature movements are substantially synchronized, in real time, to the constituent elemental speech units. In particular, a series of facial movement patterns for the constituent elemental units of acoustic speech signals, which are indicative of movement, in three-dimensions, of a number of points around the mouth of a speaker, are generated and stored, and, in response to the identification of each of the constituent elemental units of the acoustic speech signal delivered by the speaker, a respective stored movement pattern is retrieved for displaying on the display screen. The facial features around the mouth preferably include the region around the lips, cheeks and nose, and the number of points around the mouth may amount to at least ten points. |
