On determining heuristically decision threshold in robust AR speech model identification procedure based on quadratic classifier

In this work, a robust recursive procedure for identification of a nonstationary AR speech model based on a frame-based quadratic classifier with heuristically decision threshold is considered. A novel heuristically decision threshold is proposed and its efficiency is compared to the previously defined heuristically thresholds through analyzing natural speech signal with voiced and mixed excitation segments. Obtained results show that the considered robust procedure with the proposed heuristically decision threshold achieves more accurate AR speech parameter estimation and provides improved tracking performance.