Application of affine-invariant Fourier descriptors to lipreading for audio-visual speech recognition | Synapse