This thesis is about improving the television experience of visually impaired people (VIP). The company VFO, world leading in assistive technology for VIP, proposed the assignment after finding out that VIP still watch television but do not experience it like sighted users.

First, background information was retrieved by literature research. Common causes of having a visual impairment and what kinds of impairments exist were derived. Existing knowledge of television watching by VIP was still undersized; no recent study has been done. The main insight obtained from the literature research was that VIP do watch television as much as sighted people. However, VIP experience several problems while watching television that they are not yet able to solve, such as difficulties with understanding a story or missing important details. Currently some products exist to enhance the television experience of VIP but none of them fit to all visual impairments or solves their actual problem.

During this project a study was done to understand the context of VIP watching television and what problems come to light while using the television. In addition to struggles with positioning and using the television, VIP are having a hard time following the content of TV-programs. Details are missed, subtitles are too small to read or disappear too quickly and fast-changing images make it hard to focus on the subject.

In this thesis, a solution will be presented for the inability to follow a conversation because of fast-changing images and the incapability of reading subtitles. Several ideas were generated. Four concepts were formulated after discussions and brainstorms with experts of different expertises such as visual aiding products, vision of VIP and about possibilities of tracking content in a video fragment. These four concepts were tested with VIP to validate the working principle. One concept was found to be applicable for all kinds of visual impairments and a genuine help for VIP: Voices. The concept changes the single voice of spoken subtitle equipment into four voices (two male and two female). This concept needed some improvement according to VIP: making the voices more natural. Furthermore, the Dutch voices that can be used by VFO are Xander and Claire, just one male and one female voice. This evoked one more improvement that should be made: developing dual voices out of one existing voice.

Several possible voices were made with editing the existing ones. A survey was done that indicated the best voices and the needed pitch difference (two semitones) between two voices to perceive them as a different personality.

Video fragments of Dutch television with subtitles were edited with these voices to test the final concept; SubSpeech, with VIP. This last research provided the insight that two semitones distance between two voices was not always perceived as two different personalities. However, the concept was perceived as intended and participants mentioned that several voices do help them understanding a conversation due to the indication of a changing speaker. With current knowledge SubSpeech will help VIP understand conversations of foreign languages better and will make them watch television more easily. A user scenario and a brief list of recommendations for developing the concept into a product is presented at the end of the thesis.