Abstract

Language Interpreter and Speaker is a device for identifying the language of the written image text and then converting the same text to speech format. This device would surely be useful for blind and visually impaired people. Language identification (LI) is the method in which we identify the natural language of the given content. It is the process of categorizing a document on the basis of its language. In this generation, we are heading towards a phase where computers would be capable of doing all things that humans can do. Recognition of language used is the initial requirement before reading or learning. To start with any of the tasks, humans first try to understand the task and then process the task. Similarly, for language identification, the machine needs to learn the language and once learning is complete, it should be able to recognize the language. The project is divided into three parts. Initially, the handwritten image text would be converted to normal text. In the second part, the language would be identified from the converted text and last, the text would be converted to audio format. This paper discusses the implementation of this idea, gives an approach to problems and challenges that we came across, and some solutions.

Keywords

AlexNet, CNN (Convolution Neural Network), gTTS (google-text-to-speech), Image Processing