ALGORITHM OF SEGMENTATION OF A SPEECH SIGNAL BASED ON THE VALUE OF THE MINIMUM MEASURE OF DISTINCTION
A method for improving the quality of the algorithm of speech signal segmentation on voiced and unvoiced areas based on the model of the peripheral part of the human auditory system is proposed. The improvement consists in using the values of the minimum measure of distinction. The conducted research has shown: application of smoothing increases the quality of segmentation.
Keywords: speech signal, speech technology, voiced sections, minimal measure of distinction, pitch frequency, segmentation