High Impact Factor : 4.396 icon | Submit Manuscript Online icon |

Data Extraction from Web using Speech Recognition


Narendra Kumar Saini , Yagyavalkya Institute of Technology, Jaipur; Archana Mewara, Yagyavalkya Institute of Technology, Jaipur; Laxmi Narayan Balai, Yagyavalkya Institute of Technology, Jaipur


Hidden Markov Model; Mel-Frequency Cepstral Coefficient; Cross-Correlation Method


Now days many aspects of speech recognition have been taken over by a deep and smart learning method called Long short-term memory (LSTM). There are many applications of the speech recognition now a days like ticketing system, wheel chair controlling for handicapped, smart data entry, hands free computing, dictation system, smart house appliance control, authentication system, change FM channel while driving a car etc. By use of Speech recognition system any person can handle the system using their voice. The foremost objective of this paper is to extract information from web by making interface between speech and the web. The information is about the data of weather of different cities and stocks of different companies respectively. In this paper we simulate different algorithms of speech processing in MATLAB. For better recognition of speech is performed by the coefficients MFCC (Mel-frequency-Cepstral Coefficients). The efficiency of speech recognition of different words is determined in case of still voice samples. Hidden Markov Model and cross-correlation Techniques are used in this system. The isolated words are record using microphone and simulations of the system done adaptively by using MATLAB. The recognition efficiency is 65% determined by cross-correlation plotting and recognition efficiency 91.60% has been gained by using 400 samples using Hidden Markov method.

Other Details

Paper ID: IJSRDV5I110146
Published in: Volume : 5, Issue : 11
Publication Date: 01/02/2018
Page(s): 175-177

Article Preview

Download Article