Hybrid Approach for real time tricky Gujarati word recognition

Many researchers are working to make computer to understand naturally spoken language. Forinternationallanguage like English this technology has grown to a matured level. We presenta model which recognizes Gujarati spoken by human and convert it into text. The aim is recognition of the Gujarati tricky words.In this dissertation Work, I have proposed a method extracts by speech using Mel Frequency Cepstral Coefficient (MFCC) feature extraction technique and HMM model. The recognizer isworking in abundance of three essential structure squares to be specific Feature extraction,Training and Testing (Recognition). The proposed here executes the Mel Frequency Cepstral Coefficient (MFCC) with a specific end goal to figure the otherworldly components of thediscourse signal Utilizing Support Vector Machine (SVM) to perceive discourse test to give fantastic results forsecluded words. It comprises of detached words that are isolated by quiets. This proposedsystem provides high accuracy for Gujarati language