| 摘要 |
PURPOSE:To easily perform real-time processing with small deterioration in recognition rate with time by using the mean frequency characteristics of a voice as an input for speaker recognition which uses a neural network. CONSTITUTION:The input voice is passed through band-pass filters 20 of plural channels, averaged by the output averaging circuits 15 of band-pass filters 10, and inputted to the neural network 20, whose output is inputted to a decision circuit 30 and identified or matched. Consequently, the preprocessing of the input becomes simple and the neural network 20 is so structured that arithmetic processing is simple and speedy in principle and the influence of variation due to the period difference of the voice is reduced. Consequently, the deterioration in recognition rate with time is small and the real-time processing is easily performed. |
