Project In Internship:A) Information Retrieval from Non-Native Speech database using Hindi Speech Recognition : The system is used for extracting information from continuous Hindi speeches. Speech database contains a number of audio files. If some user wants such audio files which contains some specific word in it, then system is capable of giving all audio files for which user is searching. Modules:1 - Isolated Word Recognition: It is a Limited Vocabulary Isolated Word recognition system using Hidden Markov Model (HMM). HMM models are used to statically model the words in a dictionary. Dictionary contains set of words which you want to recognize. System is capable of recognizing any word for which HMM models are built.2 - Word Spotting from Continuous Hindi Speech: This is used to distinguish between Speech and Non-Speech segments using signal processing and pattern recognition and to identify word boundaries in the sentence. This module gives exact time information for every word present in continuous speech. Voiced Activity Detection and Windowing technique is used for word spotting.B)Text Area Extraction And Text Localization in Background colored Images and Optical Character Recognition This system accepts an complex colored image which containing textual part and gives out file which contains all text parts which are present in that image. This output file will help to identify the state of the image.Modules:1 - Text Area Extraction: This module is responsible from extracting text from complex colored images. User can give any complex image which has text written on image, then system extracts text part from image and gives you image, which contains only text, and background images are removed. 2 - Text Localization This module is capable of locating text boundaries in image.3 - Optical Character Recognition: This module is responsible for recognizing text from images.