University of Toronto

Technische Unversität Berlin

Jali Inc.

Eugene Fiume

Karan Singh

Pif Edwards

Chris Landreth

Christian Lessig

A system and method for triggering animated paralingual behavior from dialogue. The method including: receiving a corpus of dialogue including a plurality of samples, each sample including dialogue with aligned sequences of phonemes or sub-phonemes; extracting properties of measured quantities for each sample; generating a statistical profile by statistically classifying the extracted properties as a function of each sequence of phonemes or sub-phonemes; receiving a stream of phonemes or sub-phonemes and triggering paralingual behavior when the properties of any of the phonemes or sub-phonemes deviate from the statistical profile beyond a predetermined threshold; outputting the triggered paralingual behavior for animation.

System and method for triggering animated paralingual behavior from dialogue

A system and method for animated lip synchronization. The method includes: capturing speech input; parsing the speech input into phenomes; aligning the phonemes to the corresponding portions of the speech input; mapping the phonemes to visemes; synchronizing the visemes into viseme action units, the viseme action units comprising jaw and lip contributions for each of the phonemes; and outputting the viseme action units.

System and method for animated lip synchronization

Computer implemented method for rendering an image of a three-dimensional scene on an image plane by encoding at least a luminosity in the image plane by a luminosity function. The value of the luminosity can be computed at substantially each point of the image plane by using a set of stored input data describing the scene. The method includes constructing the luminosity function as equivalent to a first linear combination involving the functions of a first set of functions, and computing at least the value of the coefficients of the first linear combination, by solving a first linear system, obtained by using at least the functions of the first linear combination, at least a subset of the first subset of the image plane, and the luminosity at the points of said subset. The method further includes storing the value of the coefficients of the first linear combination and at least the information needed to associate each coefficient to the function multiplying said coefficient in the first linear combination. The first set of functions comprises each function of a second set of functions satisfying a selection condition, which depends at least on the set of stored input data. Moreover, the points of the first subset are distributed according to a first distribution criterion, which depends on the location of the support of at least a function of the first set of functions.

Growing community of inventors

Burnaby, Canada

Eugene Fiume