Automated read-write kit for blind using hidden Markov model and optical character recognition

In India there isn't a cost effective assistive technology for the blind students to take part in academic activities. This paper proposes a portable model kit that would convert the text in source material into speech and also converts the speech given by the user into text. The technique of Optical Character Recognition is used to convert the image captured from the source material into text format suitable for synthesis in successive stage. The converted text format is processed using Hidden Markov Model based text-to-speech synthesis to produce the equivalent speech signal. This enables the blind person to listen to the information in text format. Hidden Markov Model based speech-to-text synthesis is used to enable the blind person to respond to text information or transcribe his speech signal into textual information. The speech input from the user is converted into text using Hidden Markov Model based speech synthesis. The model kit will have the capability of automated speech-to-text and text-to-speech synthesis combined in a single unit. This unit is powered using a suitable range battery to ensure his portability.

1. Introduction

One out of every three blind people in the world live in India. An estimate of 15 million blind people lives in India. In this there are about 2 million blind children in India. Only 5% of them receive education. This figure has been largely contributed to the fact that there is no cost effect assistive technology for blind children to pursue their academic purposes. This estimation would go up in future, if there is no cost effective assistive technology for the blind. The matter of blind students in India having problems with finding writer for their competitive exams further add to this problem. The unavailability of writers for blind students and cases of writers not turning up for blind students in exam has been witnessed in the past. An assistive technology in form of automated read - write capabilities using voice control would address this problem to a great extent.