Read2Me: A Cloud- based Reading Aid for the Visually Impaired Heba Saleous, Anza Shaikh, Ragini Gupta, Assim Sagahyroon American University of Sharjah, UAE Abstract — For the visually-impaired segment of the population, the inability to read has a substantive negative impact on their quality of life. Printed text (books, magazines, menus, labels, etc.) still represents a sizable portion of the information this group needs to have unrestricted access to. Hence, developing methods by which text can be retrieved and read out loud to the blind is critical. In this work, we discuss the design and implementation of two assistive platforms, in one, we combine todays’ smartphone capabilities with the advantages offered by the rapidly growing cloud resources, and the other utilizes a more economical approach making use of cost-effective microcontrollers. Both approaches make us of an Optical character Recognition (OCR) engine on the cloud and use local resources for the Text-to-Speech (TTS) conversion. Prototypes are successfully developed and tested with favorable results. Keywords—Visual Impairment, OCR, TTS, Smartphones I. INTRODUCTION Approximately 285 million people around the globe suffer from some sort of visual disability, with 39 million being completely blind. According to the World Health Organization (WHO) [1], 1.4 million blind individuals are minors under the age of 15, and 90% of people with impairments live in low and middle income countries. Therefore, visual impairment and finding feasible solutions to reduce the burden of it is a timely issue that requires the attention of researchers in industry as well as in academia. Furthermore, Today’s technological advances provide an ideal and necessary base for finding optimal and cost effective solutions to this frustrating problem. However, despite of the entrenched research efforts in this area, the world of print information such as newspapers, books, sign boards, and menus remain mostly out of reach to visually impaired individuals. Hence, in an effort to seek an answer to this persistent problem, an assistive technology- based solution, referred to in this paper as Read2Me, is developed and tested in the work presented here. The project aims to implement a reading aid that is small- in size, lightweight, efficient in using computational resources , cost effective and of course user friendly. II. RELTED WORK To address the challenge described in the previous section, researchers have attempted to ease the burden on blind people by proposing various techniques that converts text to audible sounds. Tyflos [2] is a pair of glasses that had cameras attached to the side, earphones, and a microphone. Voice commands can be used to guide the user and direct the platform. Some commands include “move paper closer,” “move paper up,” “move paper up, right” from the device to the user, and “rewind paragraph,” “forward paragraph,” and “volume up” from the user to the device. However, the voice user interface might not function perfectly in a noisy environment, rendering it limited to indoor use. Finger Reader [3] is a wearable ring with a camera on the front. The user only needs to point at the text that they would like to read, and the Finger Reader will use local sequential text scanning in order to read each line of text progressively. This device is quite small, making it easy to carry around wherever the user goes. However, this device can give inaccurate results if aimed incorrectly, and it produces segmented audio output rather than a continuous audio string, which could confuse the user. In [4], the development of mobile applications to allow blind users to read text is discussed. OCR and TTS tools were integrated into an application in order to capture images and return audio as output back to the user. The Levenshtien Distance algorithm is used as a comparison tool between the text received after OCR has been done, and the original image. Authors experimented with three OCR tools: Tesseract, ABBYY, and Leadtools. ABBYY and Leadtools proved to be more accurate, each with approximately 18.8% of the median value of string distance while Tesseract had approximately 23.4% [4]. However, due to budget limitations for the project, Tesseract was used, since it is free, whereas Leadtools and ABBYY are commercial. The design of a microntroller- based prototype that converts images to audio is described in [5]. However, the prototype is only tested using large text images such as labels or large font titles on covers. A product Label Reader for the blind is discussed in [6]. A camera capture a video of the product then the captured video is split into frames. A text detection algorithm is then used to separate the text from sequence of frames. The OCR and TTS techniques are used to read the label back to the user. The application in [7] works by scanning the room using a wearable camera or the smartphone’s integrated camera for QR codes placed on objects around the room. The scan occurs from the left side of the room to the right, and an audio output lists objects in three different ways using AT&T’s TTS Demo. The app was tested with blind individuals to gather accurate opinions on their feelings towards the application and their uses and it received positive input. 978-1-4673-8743-9/16/$31.00 ©2016 IEEE