A Novel A2-based Neural Network Model for Voice Prediction

Abstract

A novel “Multi-Layer Perceptron” (MLP) model, based on the new A2 arithmetic is investigated in this paper. The already implemented A2 arithmetic operators, such as addition and multiplication are used to predict voice signals. One of the main techniques related to this field is the neuronal training. Unlike the static networks, where the training is done in a software manner before being prototyped, we focus ourselves on the realization of a dynamic neuronal system where the training is done continuously within the circuit. The chosen network topology will be justified later in the following paragraph. Also, the elaborated network architecture as well as a comparative study of the obtained performances compared to previous works will be the subject of the following paragraphs.

Authors and Affiliations

Hatem Boukadida| University of Sousse Higher Institute of Applied Science and Technology Sousse, Tunisia, Hassen El Fayedh| University of Sousse Higher Institute of Applied Science and Technology Sousse, Tunisia, Zied Gafsi| University of Monastir Faculty of Science of Monastir Monastir, Tunisia, Kamel Besbes| University of Monastir Faculty of Science of Monastir Monastir, Tunisia

Keywords

Related Articles

Analysis of Burst Assembly Modeling for Optical Burst Switched Network

In this paper, we have study the current state of the technology, the Optical burst Switched (OBS) network is the most practical in all-optical architecture. Here we define how Burst Assembly will carried out and also he...

Atmospheric Turbulence and Pointing Errors Effects in the Performance Matrices of Wireless Optical Links

The average channel capacity and bit error rate parameters of free space optical link under atmospheric turbulence modelled by Gamma-Gamma distribution and Kdistribution respectively with pointing error effects has bee...

Digitally Programmable Four Phase Quadrature Oscillator in Current Mode

A simple technique is presented to realize a digitally programmable current mode four phase quadrature oscillator using MOCCII. The realized oscillator enjoys the attractive features of digital tuning, low component c...

Design of Low Cost Greenhouse Monitoring using ZigBee Technology

Greenhouses are often used for growing flowers, vegetables, fruits, and tobacco plants. Most greenhouse systems still use the manual system in monitoring the temperature and humidity in the greenhouse, a lot of probl...

Assigning Subjects to Teachers Using Apriori Algorithm

Data mining is a computerized technology that uses complicated algorithms to find relationships in large data bases Extensive growth of data gives the motivation to find meaningful patterns among the huge data. Aprior...

Download PDF file
  • EP ID EP8439
  • DOI -
  • Views 354
  • Downloads 22

How To Cite

Hatem Boukadida, Hassen El Fayedh, Zied Gafsi, Kamel Besbes (2015). A Novel A2-based Neural Network Model for Voice Prediction. International Journal of Electronics Communication and Computer Technology, 5(3), 848-853. https://europub.co.uk./articles/-A-8439