The legal domain often deals with large documents, witness testimonies, and multilingual content, which is time-consuming and resource-intensive. This research presents Smart Justice, an AI-driven system designed to automate key aspects of the legal process, consisting of three integrated models. The Speech-to-Text Witness Translator accurately transcribes Hindi audio testimonies into English using Google Speech Recognition and Google Translator API, with a 90% accuracy. Judgement Summarization model is a tool that efficiently processes Supreme Court judgment documents using the BART-based summarization model, with an accuracy of 92%. Finally, Document Translation model translates the English legal document to Hindi language using the MBART-50 multilingual translation model with an accuracy of 90%. Integrating Natural Language Processing and machine learning techniques in a legal proceeding smoothens the work of document analysis, multilingual communication, audio evidence processing, reduces manual efforts, improves accuracy, and helps non-native speakers access it as well, thereby making it one of the finest tools for legal workflow modernization.