On Voice Activated Information Retrieval System

Speech Recognition (SR) is a process that transcribes speech into text using a computer. Speech recognition system is a speech-to-text conversion wherein the output of the system displays text corresponding to the recognized speech. A step towards a more natural, “human-like” communication between machines and users in need of information is represented by the introduction of speech language technologies into Information Retrieval system. The integration of the Information Retrieval system (IR) and the Automatic speech recognition (ASR) system degrades the performance. The failure to recognize a keyword in continuous speech may drastically affect the performance of the IR system. The proposal is to build an ASR system that is trained to identify the word we pronounce in restricted domain. Then for each of the recognized word, the system is expected to find the match which is then extracted by the IR system.