Design of Simulator for Automatic Voice Signal Detection and Compression (AVSDC)
Shiv Kumar1, Aditya Shastri2
1Shiv Kumar, Research Scholar, Department of Computer Science & Engineering, Banasthali University, Tonk (Rajasthan), India.
2Prof. (Dr.) Aditya Shastri, Vice Chancellor Banasthali University, Tonk (Rajasthan), India.
Manuscript received on March 01, 2012. | Revised Manuscript received on March 04, 2012. | Manuscript published on March 05, 2012. | PP: 10-38 | Volume-2 Issue-1, March 2012. | Retrieval Number: A0485032212/2012©BEIESP
Open Access | Ethics and Policies | Cite | Mendeley
© The Authors. Published By: Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (

Abstract: A good amount of work has been done in the field of compression, voice signal detection, and spectrum analysis which has been generated a number of results in the past few decades. In this research, following three important problems have been identified: 1. To distinguish between constitutional and unconstitutional Voice: It is an important task to identify authenticity of recorded voice of the specific person. Here it has been tried to develop a Simulator which identifies constitutional and unconstitutional voice. 2. To identify words sequence:It is an important task to recognize words sequence in the recorded voice. Sometimes voice may be recorded fast, clear, or loud. Here it has been tried to develop a simulator to checkout whether recorded words are in proper sequence are not. 3. To develop a simulator which does not change file extension and quality of voice signal after compression: Normally, after compression, file extension is changed and quality of the voice signal is deteriorated. Here it has been tried not to change extension of the file after compression with minor distortion in voice signal. As per review of above three problems, it is being considered a simulator may be designed which may resolve above problems. With this view, the research title is chosen as “Design of Simulator for Automatic Voice Signal Detection and Compression (AVSDC)” which is suitable for pervasive computing, voice signal detection, and spectrum analysis. AVSDC is divided into following two parts: 1. Automatic Voice Signal Detection (AVSD) 2. Automatic Voice Signal Compression (AVSC) Automatic Voice Signal Detection (AVSD) is used to identify constitutional and unconstitutional voice signal automatically which is performed on the basis of frequency, pitch value, formant value, and sequence of words in the voice signal for several samples of the same voice. An underline purpose of AVSD is to identify fake voice in the security system. Frequency is being mapped to the frequency domain by computing its DFT using the FFT algorithm. Sequence of words is computed by continuously computing difference between absolute averages of two adjacent significant windows and comparing it to a predefined threshold. Word Identification System is part of AVSD which is designed to checkout whether recorded words in proper sequence are not. Normally, sometimes spoken words of voice may be recorded very fast, smoothly, or loudly. The main idea behind the word identification system is to first train it with several versions of the same word, thus yielding a “reference fingerprint”. Then, subsequent words can be identified based on how close they areto this fingerprint. The whole idea is evaluated on the basis of Euclidean distance theory. Automatic Voice Signal Compression (AVSC) takes .wav stereo file as an input and compress 50 to 60 percent of the source file at about 45 kbps with high quality voice signal by taking the help of adaptive wavelet packet decomposition and psychoacoustic model. AVSC takes .wav stereo file as an input and creates .wav mono file after compression. After compression minor distortion is also possible. The main feature of AVSC is that file extension does not change after compression. In other words, compression is done from .wav to .wav extension. AVSC takes .wav stereo file as an input and after compression it creates .wav mono file as an output. AVSC also computes entropy and SNR (Signal to Noise Ratio) of the source file during the compression.
Keywords: MatLab7.0, Euclidean Distance Theory, Wavelet, Frequency Volue, Pitch Value, Average Significant Window