The vast information available in the web data is mostly inaccessible for blind and visually impaired since most information retrieval systems do not cater to their needs. The retrieval of the relevant information from multiple web documents in a synoptic form can be highly beneficial. A system with speech user interface for information retrieval of web-data in summarized form was developed. The system is speaker-independent and uses a domain specific language model. Query is formulated based on user's input which can be either in text or in voice form. The query is processed by a search engine, like Google, and a fixed number of relevant web documents are parsed and the textual content is extracted from them. This textual content is summarized and its corresponding speech along with the text is presented to the user. Through this method an enhanced solution which offers a natural and user-friendly way of communication for visually-impaired people is intended. The system is easy to be implemented in real time and its functioning is simple as compared to other existing systems.