Acoustic Feature Research of Snoring Signal Based on Time Frequency Analysis

  • 王 恆康

Student thesis: Master's Thesis


Currently about one-third of the world population suffers from sleep disorders and as such the quality of sleep has been an active topic of research Sleep apnea syndrome is one of these sleep disorders and obstructive sleep apnea syndrome (OSAS) has dominated due to the 90 percent such a high percentage compared to other types among it Sleep apnea patients often suffer from snoring sleep disorganization continual oxygen desaturation daytime sleepiness impaired concentration and various other symptoms Moreover OSAS may be related to cardiovascular disease Currently polysomnography (PSG) is the primary method to study OSAS Data derived from monitoring the sleep of patients overnight with PSG is the main basis for diagnosing OSAS and allows patient assessment and the selection of appropriate therapeutic measures However PSG is very costly and its leading signals have high multiplicity and complexity During the procedure patients may have difficulty sleeping due to being unaccustomed to the measurement apparatus or they may feel uncomfortable which may lead to deviation Additionally patients need to stay in the hospital overnight Due to the fact that the procedure is time consuming and since there is often a shortage of beds PSG is not suitable for widespread use Therefore the main aim of our study is to construct an accurate economic simple and portable OSAS diagnosis system that can overcome the limitations of PSG such as mass screening of OSAS Previous research has reported a system of recording and analyzing the electrocardiogram (ECG) and oxygen saturation (SpO2) levels In this study our major work is the acoustic analysis of snoring signals recorded by microphone We used MATLAB to extract features of collected snoring signals and then analyzed features containing frequency distribution power ratio 800 (PR800) formant frequency and Mel frequency cepstrum coefficients (MFCCs) which can provide the basis of construction of our OSAS detection system
Date of Award2016 Aug 25
Original languageEnglish
SupervisorTainsong Chen (Supervisor)

Cite this