Enhancement of Speech Signal Segmentation Using Teager Energy Operator

Abstract

The article presents the improved method for speech signal segmentation, which provides an increase in the efficiency of detecting voiced and unvoiced sections, and pauses using the Teager energy operator. The method is based on energy analysis of speech signal fragments using the Teager energy operator with the subsequent analysis of zero-crossing rate and short-term energy of the energy characteristic function. The research was carried out to assess the performance and noise robustness of the proposed method in comparison with methods based on the analysis of zero-crossing rate, short-term energy, and one-dimensional Mahalanobis distance. In accordance with the obtained research results, it was concluded that due to the good susceptibility of the Teager energy operator to changes in signal amplitude and frequency, the improved method provides an enhancement of segmentation of speech signals, including noisy ones. Depending on the requirements for segmentation accuracy, the proposed method provides variability in the values of the first and second kind errors by changing the threshold coefficient.