Fused HMM-Adaptation of Synchronous HMMs for Audio-Visual Speech Recognition