Acoustical analysis of the cause of snoring noise

Introduction: For a targeted surgical therapy of the obstructive sleep apnea syndrome (OSAS), it can be of benefit to know where snoring is generated within the upper airways. Today, drug induced sleep endoscopy (DISE) is increasingly used to identify location and form of vibrations and obstructions. However, DISE is time consuming and costly. In contrast, determination of the snoring noise location by means of acoustic analysis could be easier for doctor and patient. Furthermore, it can be performed during natural sleep. Aim of this study is to evaluate a method for acoustical analysis of snoring noise, which is based on principles of speech signal processing. Methods: Up to now, 25 patients with primary snoring and OSAS, who underwent DISE, have been included in the study. Using DISE video and audio recordings, the vibration locations where classified based on the VOTE classification (velum, tonsils (oropharynx), tongue base, epiglottis). The snoring events where extracted from the audio data. By means of speech signal processing methods (LPC, MFCC), the frequency response of the articulation tract was calculated in order to find snoring formants which are characteristic for the snoring sound location. Results: Based on the first and the second formant, characteristic frequency patterns could be distinguished for the vibration locations epiglottis, tongue base, and oropharynx/velum. These patterns are independent of the source signal9s fundamental frequency. Conclusion: In the future it may be possible to detect the location of snoring and obstructions within the upper airway by acoustical analysis of snoring during natural sleep, in order to support the selection of an appropriate therapy for the patient.