Development Of Speech Recognition System For Forensic Application

You, Kean Heng (2014) Development Of Speech Recognition System For Forensic Application. Project Report. UTeM, Melaka, Malaysia. (Submitted)

Text (24 Pages) Development Of Speech Recognition System For Forensic Application 24 Pages.pdf - Submitted Version Download (458kB)

Abstract

Speech Recognition is the process of recognizing the speech spoken by a particular speaker based on important feature of speech waveform. Forensics is the use of science or technology in the investigation and establishment of evidence in the court of law. This process involves the comparison of recordings of an unknown voice with one or more recording of a known voice. However, there have variability in individual’s speech and so the optimum technique for the speech recognition system is yet to be decided through the vast efforts of researchers. Signal processing front end is the process for extracting the data feature in speech recognition system. There have techniques of signal processing front-end such as Linear Prediction Coding (LPC), Mel Frequency Cepstrum Coefficient (MFCC), and other. The state-of-the-art in feature matching techniques used in speech recognition includes Dynamic Time Warping (DTW), Hidden Markov Modelling (HMM), and Vector Quantization (VQ). The objective is designing a speech recognition model using MFCC extraction technique and also with Vector Quantization model. The voice is based on isolated or single word recognition. In this project, there are 8 persons providing 5 speeches of each person for database. Every person utter twice for each speech as database. These files are recorded in Microsoft WAV format. The algorithm has been developed and the speech waveform has been analysed. The accurate and reliable speech recognition system has been developed for forensic application.

Actions (login required)

View Item