Nigerian Language Simulated Speaker Verification System Using Back-Propagation Neural Network

Abstract

This paper presents the use of Back propagation neural network to implement speaker verification simulated with Nigeria Languages. The focus is to verify voice patterns for different people speaking different Nigeria languagesso as to recognize (verify) their speech electronically.The voice samples of the people utilized were captured and then processed using the sound forge 9.0 software.The frequencies of each voice signals were used to train a backpropagation neural network, which inturn verifies the speakerthrough the voice patterns. Six neural networks (K1-1, K2-1, Y1-1, Y2-1, F1-1, and F2-1)were developed for training, testing, and validation ofthree selected Nigeria languages words from nine (9) people (three for each language). Discrete Fourier Transform (DFT) was usedto extract features from the voice samples for the backpropagation neural network (BPNN) training, testing, and validation. The network’s performance analysis results as deduced from regression analysis showsan averageoverall R-value of 0.9371 for acceptance, and an average of 0.3414 for rejection. The results obtained shows that each network verified the speaker it was trained for adequately. The results obtained from this work can be generalized to cater for larger vocabularies and for continuous speaker verification processes.

Authors and Affiliations

Shakiru Olajide KASSIM, Mohammed ABDUL- FATAU

Keywords

Related Articles

Impact of Large Scale Renewable Integration in Bangladesh Power System

Bangladesh, a small developing country of south Asia observing a rapid economic transition from agriculture to industry and facing a threat of unsatisfied consumer demand of energy exacerbating with growing population. K...

Construction of a Continuous Flow Eletroflocculation Prototype Supplied By Photovoltaic Energy for Effluents Purification

Effluents production in the most varied industrial sectors has favored environmental degradation, especially when not treated and disposed in soil or water sources, for this reason, this work main objective was to design...

Projected Load Growth Sustenance Investigations by Modelling and Simulations of Electric Power Distribution Networks: A Case Study of Sekondi-Takoradi Metropolis

The reliability of electric power distribution networks has gained greater interest in recent times. One of the main causes of frequent failure of electric power distribution networks is exceeding the current carrying ca...

An Approach Using Sudden Change in Geometrical Shape of a Sample for Fabricating Micro Materials Utilizing Electromigration

The utilization of electromigration of induced micro materials in passivated aluminum films was investigated. The experimental sample was a passivated polycrystalline Al line with a hole using sudden change in geometrica...

Effect of Cell Size on the Population, Average Call Time, Subscriber Number Per Call(SNPC) And Call Number Per Hour

When we planning for cell mobile area, we should put in our consider the traffic and the coverage area by doing analysis using the TDMA – FDMA mobile cellular communication system. In this paper we collaborates with Oore...

Download PDF file
  • EP ID EP402411
  • DOI 10.9790/1676-1305016270.
  • Views 164
  • Downloads 0

How To Cite

Shakiru Olajide KASSIM, Mohammed ABDUL- FATAU (2018). Nigerian Language Simulated Speaker Verification System Using Back-Propagation Neural Network. IOSR Journals (IOSR Journal of Electrical and Electronics Engineering), 13(5), 62-70. https://europub.co.uk./articles/-A-402411