Speaker identification in time-sequential images based on movements of lips

The extraction of biological symbol feature decided the recognition effect in speaker identification. Because the traditional feature extraction of lips based on single static image neglected the movement of lips in speaking, a method of speaker identification in time-sequential images based on movements of lips was proposed. Firstly, feature information of each frame in time-sequential of videos was extracted. Secondly, HMM models of movements of lips of speakers were constructed and trained. Finally, compared probability between observed symbol sequence and HMM models of database to identify the speaker. The experiments indicated that the algorithm was well done in feasibility and effectiveness and had good recognition effect.