An automated text to speech conversion system a special reference to Assamese language

Abstract

xv newlineAn Automated Text-To-Speech Conversion System: newlineA Special Reference to Assamese Language newlineABSTRACT newlineNatural Language Processing (NLP) is a new field of study and research in computational linguistic. It combines machine learning, computational linguistic, and deep learning models to process human language which is incredibly diverse and may be spoken in many different ways. Since NLP deals with human natural language, the main difficulties are handling ambiguity, representing common senses, retrieving contextual information, pragmatic analysis, etc. The two main focuses of NLP research are speech recognition and text analysis. At present scenario, text to speech conversion is one of the important activities under NLP paradigm. Understanding human languages is a challenging task for a machine. It requires regorious training to develop a trained model. Due to the complexity of a language, like countless nuances, dialects, and regional variations, it leads much work to standardize the model. The latest breakthrough in natural language processing is Text-to-Speech (TTS) conversion. It converts written data into audio files with excellent speech quality. With the developed system people who are unable to read or visually impaired, now and onwards can read written text such as article, newsletter, or any other documents available in softcopy. With this system, people can read text when at work or at driving. Many systems have been developed and in used but they are in English language. Till date there is a need to develop similar system for each regional languages and also that s why there is a demand to develop a text to speech conversion system for regional language like Assamese. Assamese is a regional natural language spoken by the people of entire north-eastern region of India. newlineIn text to speech conversion many approaches are used like Dictionary based, Hidden Markov Model, Convolutional Neural Network and Grapheme to phoneme. In Dictionary based approach, a database is created by recording word initially. For a