Enhancing speech recognition in developing language learning systems for low cost Androids

Learning to read correctly is a key requirement of language learning. In rural India, due to lack of teachers and technology, tablets offer a creative and motivating learning environment. Tablet technology has the advantage of mobility, allowing users to learn at their own pace and convenience. However, the non-availability of electricity and Internet can be unique challenges. At Amrita CREATE, language-learning solutions have been developed for students to learn and read on the tablets. It uses advanced speech recognition technique to provide feedback and intervention. Proposed system is unique in its ability to evaluate words and phrases and corrects the learner as they articulate the sentence. This system works without Internet and on the lower processing power of android tablets. Silence detection and multiple-synchronized recognition have been introduced in this paper which greatly enhance the ability to provide feedback to the user in real-time. The combination of the two helps in achieving successful recognition of longer and continuous sentence.