Towards realization of an ANN model of the linear prediction coder
M.P. Beddoes · 2003
Two forms of artificial neural networks, ANNs, each a possible substitute for the linear prediction coder, are studied. The application is in the field of speech processing. We show experimentally that ANNs behave in much the same way as an LP coder. However, ANNs were trained using the backpropagation algorithm, and this algorithm may avoid ambiguities which can and do occur with the original LP coder.