Cost Efficient Bangla Book Reader for the Visually Impaired

The process of converting physical books or paper documents to a digital format is commonly known as book digitization. In this paper, we propose a cost-effective Bangla book reader for the visually impaired people with the help of Raspberry Pi. A successful model has been developed which scans a page from a physical book, identifies the text using OCR technique, translates needed segments into Bangla using Google Translate API, reads the text aloud using TTS engine and in the process creates a digitized version of the provided book. An external webcam is attached to the Raspberry Pi to take pictures from a given book, after processing the taken images are transformed to text using Tesseract Optical Character Recognizer (OCR). The parts that are not in Bangla are translated accordingly by the Google Translator API and the processed text is transformed to audio by eSpeak NG Text-To-Speech (TTS) Engine. The audios are read aloud and also saved page by page to be combined later to create a complete audio book. Using the scanned pages of the given book that was prepared for the OCR, a PDF version is also prepared. The complete process has automatic page turning mechanism implemented at a hardware level to make it spontaneous.