Anvitha Infotech
Audio Speech to Sign Language Converter is a Django-based web application that converts spoken or typed text into Indian Sign Language animations. Users can sign up, log in, enter text manually, or use the microphone …
The Audio Speech to Sign Language Converter is a web-based application designed to reduce the communication gap between hearing users and people who rely on sign language. Conventional communication tools often lack real-time, accessible mechanisms for converting spoken or written language into a visual format that sign language users can readily understand. This system addresses that gap by accepting spoken input through a browser microphone or typed text from the user, processing the sentence, extracting meaningful keywords, and displaying the corresponding Indian Sign Language (ISL) animation clips, enabling a direct visual translation of everyday communication.
The project is developed using the Django web framework for backend routing, authentication, and page rendering, providing a structured and secure foundation for managing user interactions. The frontend uses HTML, CSS, and JavaScript, while browser-based speech recognition is handled through the Web Speech API, allowing spoken input to be captured and converted into text directly within the browser without requiring additional external speech-processing infrastructure.
Natural Language Toolkit (NLTK) is used for text preprocessing tasks such as tokenization, part-of-speech tagging, stop-word removal, tense detection, and lemmatization. These preprocessing steps refine the raw input sentence into a simplified, grammatically structured form that aligns more closely with how sign language conveys meaning, improving the accuracy and relevance of the resulting animation output. The application also supports translation of entered text into English before processing, which helps users provide input in different supported languages, broadening accessibility for a linguistically diverse user base.
After preprocessing, the system checks whether an animation clip is available for each processed word. If a complete word animation is not available, the word is broken into individual letters and the corresponding alphabet animations are displayed. This fallback approach ensures that the system can still provide visual output even when the exact word is not present in the animation dataset, maintaining communication continuity rather than failing silently when encountering unfamiliar vocabulary.The application includes user registration, login, logout, static animation assets, an interactive converter page, and informational pages such as Home, About, and Contact, rounding out the platform into a complete and usable web application. Overall, this project demonstrates how speech recognition, natural language processing, and web technologies can be combined to create an accessible communication aid using sign language animations.