A neural-network-based system for testing speakers
M.J. Er, T.H. Ooi, C.T. Toh, F.S. Toh · 2002
The paper presents a high performance neural network based system for testing speakers. A multilayer neural network system with a backpropagation learning algorithm is employed. It consists of 53 input nodes, one hidden layer with 10 nodes and 1 output node. The normalized total harmonics distortion (THD) values of the speakers at different frequencies are fed to the input of the system. The average training time is 40 minutes (on a 486DX 50 MHz PC) for a training size of 100 patterns. The neural network based system is able to achieve a remarkable accuracy of 95%.>