Enhanced captioning - Speaker identification: text vs. images

The enhanced captioning system was developed using a participatory design methodology to improve speaker identification using avatars. These avatars consisted of the speaker's name and image surrounded by a coloured border matching their clothing. An evaluation was conducted using eye-tracking with viewers who are deaf and hard-or-hearing. The major findings indicate that using an avatar image as a method for identifying who is speaking is not only feasible but may require the least attention from viewers to process. Further, differences between hard-of-hearing and deaf participants are reported.