A new approach using hidden Markov model for generating movie captions to aid the hearing impaired, which is capable of visual encoding emotional information

Movie captions comprised of pure text or some restricted selection of symbols, as produced today for movies and television are, by themselves are not capable of conveying to the hearing impaired valuable information encoded into the sound-track, namely, dialogue tone, music aliveness and background sounds atmosphere. This drastically reduces the capacity of the hearing impaired to have a better understanding and appreciation on what is truly happening in the program or movie, he is currently watching. The present paper introduces a new approach that affords captions with the expression power of conveying to the hearing impaired a great deal of the lost emotional content from the sound-tracks. Some experimental examples are provided and also commented upon.