Pitch analysis in snoring signals from simple snorers and patients with obstructive sleep apnea

Abstract

The relationship between snoring and obstructive sleep apnea syndrome (OSAS) has been reported in the literature. Classically, sound intensity and spectral features were the parameters measured for each snoring episode. Recently, studies of pitch analysis in snoring signals have also been published. Availability of automatic detectors of snores, and variability of snoring signal during polysomnography, are the main limitations to analyze the pitch. In this work, we applied an automatic snore detector previously developed by our group. The main goal of this paper is the pitch analysis of snoring signals. For each snoring episode, we calculated different parameters to quantify the pitch: mean value, standard deviation and density of pitch. A total number of 447 inspiratory snores from 8 normal subjects and 236 post-apneic snores from 8 OSAS patients were analyzed, taken at random from a database of 30 snorer subjects with different apnea/hipoapnea index (AHI). Results show a relationship between pitch and snoring patterns. Differences in pitch parameters were found between simple snorers and OSAS patients. A relationship between pitch parameters and AHI is also shown. Pitch analysis of snoring signals provides a promising tool for monitoring snoring activity and their relation with OSAS pathology.