Pitch detection with average magnitude difference function using adaptive threshold algorithm for estimating shimmer and jitter

In order to estimate jitter and shimmer, we suggest a method for the detection of precise pitch and amplitude of voiced speech signal. After low-pass filtering, voiced/unvoiced decision and pitch detection is performed using the average magnitude difference function. We find a minimum value at t=/spl tau/ and also check the possible minimum value at t=/spl tau//2, /spl tau//3, /spl tau//4, /spl tau//5, /spl tau//6. By the application of adaptive threshold, we can estimate pitch more accurately.