Enhanced Synthesized Text Reader for Visually Impaired Users

In this paper we propose a prototype of a spatialized text reader for visually impaired users. The basic functions of the system are reading arbitrary files, converting text into speech using different synthesized voices and spatializing synthesized speech. Visually impaired users can thus listen to the content of a file read by various synthesized voices at different spatial positions. Some metadata (e.g. pre-inserted tags) has to be added to the file before processing in order to define the voice, pitch, reading rate and originating spatial position for any part of the content. We believe such a way of electronic book reading can be a significant improvement for visually impaired users if compared to mundane and dull screen readers. The core of the system is based on Java platform using FreeTTS speech synthesizer and JOAL positioning library. The latter is improved by the external MIT Head Related Impulse Response (HRIR) library. The use of headphones is obligatory in order to perceive spatial sound correctly. The system is a work in progress and is currently under evaluation by twelve visually impaired test subjects.