Estimation of Electric Shaver Sound Quality using Artificial Neural Networks

For a competitive market, sound quality is an important attribute of a particular product that effects end user preferences which discriminates rival products. Traditional sound quality assessments include subjective jury testing which is both expensive and time consuming. However, contemporary studies in that field mostly focus on development new methodologies that can replace subjective jury testing, mainly artificial neural networks (ANNs). Main reason underlying this search is to compensate the shortcomings of jury testing.In this study, artificial neural networks are used to predict subjective pleasantness estimations of electric shaver sounds. Psychoacoustic parameters obtained from different shaver recordings by using different psychoacoustical models and subjective annoyance estimations gathered from jury testing are used as inputs and outputs of a neural network, respectively. With the correlations between input and output values obtained from jury testing, neural networks are trained and correlation rates are evaluated. Some of the sound samples are used for training the neural network while the rest of the data are used to verify the accuracy of the artificial neural network. Conclusive remarks are included at the end of the study related to idea of replacing the jury testing with neural networks with possible strengths and shortcomings.