Day by day, the use of the internet is increasing; as a result, the way of communication through the internet is growing dramatically. So, speech signal compression is being a vital issue in the area of signal processing. There are many techniques used to compress a speech; such as Huffman coding, LZW coding, Arithmetic coding etc. These techniques convert a speech signal in such a way so that it takes very less memory and can be sent quickly over the internet than original data. From the point of view, this article introduces a novel speech coding method which use a set of binary combinations as needed to encode speech data. In this technique, the binary combinations are set to the probabilities of a speech signal where 0 and 1 will be assigned to the first two highest probabilities. The third, fourth, fifth, sixth, seventh and eighth probabilities are assigned by 00,01,10,11,000,001 and the process will continue until all probabilities are finished. Here, Average code length (ACL), compression ratio (CR), compressed file size and PSNR are considered to evaluate the performance of the proposed technique with the state of the art techniques. From the experimental outcomes, we have seen that the proposed method provides a better result than the state of the art procedures.

A lossless speech signal compression technique

In this paper, the proportionate normalized least mean square (PNLMS) and its variants, µ-law PNLMS (MPNLMS), improved PNLMS (IPNLMS), and filter PNLMS (FPNLMS) algorithms are simulated for different types of input signals for a sparse system identification problem. The input signals such as a Gaussian random signal, band-limited signal, a speech signal, a color signal, and a uniform random signal are analyzed in this paper. The input is compressed by a wavelet transform before application. The simulations are carried out to examine the behavior of Gaussian noise on a sparse physical system. The simulation further justifies the robustness of the FPNLMS algorithm to the change in the input signal and noise level.

Wavelet Transform for Signal Compression in Sparse Algorithms

Forward Error Correction (FEC) is a digital signal processing technique used to increase the reliability of data transmitted over noisy communication channels and to provide error control. Error checking is performed using residual data, which is called error correction code before data transmission. Cyclic coding is one of the FEC techniques used to solve messages properly in modern wireless communication systems. Wavelet Transform is also used extensively in audio, image and video processing applications. In the first part of this study, the Cyclic Coding Method was used to transmit the audio data recorded in .wav format for 5 seconds in MATLAB environment with the least possible error. Binary Phase Shift Keying (BPSK) modulation is used as the modulation method. In the generated communication system, Additive White Gaussian Noise (AWGN) channel is coded with Cyclic (7,4), Cyclic (15,7), Cyclic (17,8) and Cyclic (31,21) codes respectively. The Mean Square Error (MSE) performances of these codes were then analyzed. In the second part of the study, it was researched which wavelet family is suitable for audio transmission with minimum error. Four wavelet transform families of different levels are applied to the audio signal for Cyclic (7,4), Cyclic (15,7), Cyclic (17,8) coded channels and their performances are compared.

Cyclic Coding and Discrete Wavelet Transform Based Audio Signal Processing

In this study, the audio transmission performance of some block coding methods in wavelet and non-wavelet cases is compared. For this purpose, a communication system has been established in the MATLAB environment for the transmission of the audio signal for 5 s. The comparison of the transmission performance of the block codes was performed separately in different parameters of additive white Gaussian noise, Rician and Rayleigh channels. At the same time, it is determined which discrete wavelet transform (DWT) family is suitable in terms of transmission performance for block coding methods and how DWT level affects transmission performance is investigated. Four DWT families at different levels were applied to the audio signal for linear block coding, Hamming, Cyclic, Bose–Chaudhuri–Hocquenghem, Reed–Solomon, encoded channels, and bit error rate and mean squared error performances were compared.

https://ietresearch.onlinelibrary.wiley.com/doi/epdf/10.1049/iet-com.2019.1114

Comparison of communication performance of block coding methods in different channels using DWT

The development in the field of digital communication has led to drastic improvements in the way human speech is handled for transmission. Voice communication has become a booming field for a few decades. Speech compression is used to eliminate the redundancies present in the speech signal and is associated with its efficient transmission. Speech compression is achieved by using Speech Coding. Speech coding uses speech-specific parameter estimation utilising digital signal processing techniques on audio signals, to model the speech signal. The model obtained is combined with generic data compression algorithms to represent the resulting modeled parameters in a compact bitstream.

http://ieeexplore.ieee.org/iel7/6881541/6892507/06892538.pdf

Jaydipta

Papers

Wavelet based Speech Coding technique using median function thresholding