Dynamic Audio-Visual Speaker Identification using Lip Model and Multi-Stream Hidden Markov Chains

Biometry is the science of identification of human by using physical characteristics. Use of dynamic Audio-Visual features has the advantage of improving robustness to the environmental noise and relatively lower cost than other Biometry methods. In this paper we have proposed a dynamic lip model system to extract the intrinsic parameters of moving limbs of face. these parameters are exclusively dependent on the neuromuscular properties speak. so imitation or false acceptance could be reduced to a large extent. False rejection have been increased up to0. 5 percent and true acceptance has been increased up to 97 percent in low signal to noise ratios as 6 decibel

کلیدواژه ها:

speaker recognition ، Markov chain ، lip movement