Hardik Srivastava

Style can be thought of as a way where semantics are represented intuitively. Every person has a unique writing style, which can be expressed through certain stylistic features present in a sentence. These features can include the usage of certain common words, contractions, metaphors, slang, and some syntactic structures that when combined determine the language style of a person. Currently, there is no way of applying stylistic variations on text in a dialogue system to adapt to a user or an audience. In this paper, we propose a network architecture to transform machine-like robotic conversations into human-like personalized ones by applying custom text styles to the responses thereby giving them a user-specific tonality, leveraging Language Modeling, Generation, and Style Transfer techniques in a Representation Learning fashion. When executing the process of transferring the text style, our method keeps the original sentence's content intact in the transferred sentence dissociated from its style. This system implementation can be transformed into a software application that can help people use the language styles of their acquaintances and make the AI talk like them.

According to the WHO, around one billion people suffer from disorders in speaking and experience motor-skill challenges which can often lead to exclusion from the society. An enhanced and accelerated device for communicating is something the disabled people cannot afford, who still form the world's largest minority to experience discrimination. So, in this paper I would like to propose an application, HearMeOut that uses artificial intelligence techniques to implement state-of-theart algorithms and Natural Language Processing for predicting meaningful and appropriate text, generating sentences using pictograms and facilitating short communication using speech recognition. These specifications are delivered to the user via a language assistant called LGenie (Language Generation through Input Event). This will help reduce the communication barrier between the speechless and their interlocutors thus allowing them to integrate into the existing social structures and make the world a more inclusive place.