Establishing document creation support for people with upper limb disabilities by hands-free speech recognition and gaze detection

Abstract

Everyday life is now inundated with the need to send, receive and process information through the written word. Many workplaces rely heavily on email for communication between colleagues, both local and across the world. News sites, social media and even remaining in contact with friends all primarily employ text as the language medium of choice. Naturally, there are many reasons for this, ranging from the convenience of giving someone time to think and reply to keeping information more private. For the able-bodied, typing on a computer or smartphone is quick and easy. However, these typing-based modes of interaction are much more difficult for those who have upper limb disabilities. If you cannot use your arms or hands in the same manner as the able-bodied, you cannot easily participate in this world of text-based communication. Better solutions are required. In recent years, technology aimed at converting speech into text has been growing. Such functions now exist on all smartphones and many computers, as well as in other mediums like digital personal assistants. This sort of technology can be very convenient if you have a lot on your hands, but it can be transformative for those with upper limb disabilities. The problem, however, is that speech recognition technology still struggles in a few key areas. Firstly, when there is background noise for a microphone to detect, this can confuse the speech recognition software. Background noise, of course, is nearly always present through everyday occurrences such as traffic, other people's conversations and reverberations. The second major problem is the recognition by the software of punctuation. Both cases lead to errors in transcription that are usually corrected manually by the user. Naturally, this is not possible for the upper limb disabled. Finding better solutions to the current shortcomings of speech-to-text software is a focus for Assistant Professor Ryoichi Miyazaki of the Department of Computer Science and Electronic Engineering at the National Institute of Technology, Tokuyama College, Japan.