Non-visual object generation model to ease music notation script access for visually impaired

摘要

Music scores with notations and lyrics are considered as the main mediator in musical communication channel which lies between a composer and a performer. Music learners with visual impairment have less opportunity to access this main mediator since most of them are in visual format. With the assistance of a third-party, they convert visual music notations into Music Braille, again a complex format having spatio-tactile symbols. The research tries to redefine the best mediator in composer - performer musical communication channel when at least one of the participants is visually impaired. Music and Singing can be described more precisely using a format in temporal domain compared to a visual format such as a notation and lyric script in the spatial domain. The research proposes a non-visual object generation model specifically based on synthesized singing together with a controllable interface which can convey all type of information enclosed within a music notation script. Calculated scores on similarity and user scores on naturalness relevant to the synthesized auditory output reveal that this approach can be successfully utilized to fill the accessibility gap prevailed in assistive technologies for visually impaired.