Indian Sign language provides an important means of communication for deaf and hard of hearing individuals in India. However, communication might be difficult when people who do not understand ISL have to interact with Sign language users. Artificial Intelligence, Computer Vision and deep learning offer possible ways for recognizing signs from visual inputs and converting recognized information into text and speech. This paper reviews existing research on ISL recognition, translation, datasets and sign to speech conversion. It also discusses some techniques such as Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM) Networks, Gated Recurrent Units (GRUs), and transformer-based models. It also examines datasets, such as INCLUDE and ISL Translate, alongside international sign language datasets used in related research work. The reviewed literature shows the distinction between recognizing individual signs and translating continuous sign sequences into full sentences. Some important challenges include limited and diverse ISL data, changes in the variation among signers, changes in lighting and background, and the need for effective real time processing. Based on the reviewed studies, the paper shows a conceptual framework that combines sign recognition, language processing, and text to speech conversion. The framework provides a direction for further future research; its complete implementation and performance evaluation remain outside the scope of this review.