Various Methods for Visual Speaker Identification for Automatic Continuous Speech Recognition in TV Broadcast Programs

This paper is about different methods and algorithms that were used for speaker identification from the video recordings of TV broadcast news transcription. The information from visual speaker identification were used in our complex system for automatic continuous speech recognition of TV broadcast programs because it is possible to use speaker adapted (SA) Hidden Markov Models (HMMs) if we have the information about a speaker and the resulting recognition rate is higher than if only the speaker independent (SI) HMMs are used. These methods for visual speaker identification were evaluated according to the highest recognition rate in this work.