Effect of tiredness on voice signals using neural network systems

We investigated the effect of tiredness produced by a Kraepelin test on voice signals using neural network systems for an individual identification. From the results, we found that the difference between the flicker values before and after the Kraepelin test was related with tiredness. The output values of the neural network could be classified into voice signals before and after the Kraepelin test. As a result, the higher correction rates more than 90 % were obtained and the voice signals were classified into the patterns with and without tiredness.