ArticlePDF Available

Abstract and Figures

In recent years, several studies have shown the relationship between snoring and obstructive sleep apnea syndrome (OSAS). Besides time domain analysis of snoring signal, the spectral features and shapes of snores can be used to discriminate simple snorers and OSAS patients. In this study, we propose a method to classify simple snorers and OSAS patients based on spectral envelope estimation of snoring signals. The formant frequencies and corresponding bandwidths are computed for both group, and the variation and consistency of the formant distributions are computed. A total of 1400 snoring episodes from 7 simple snorer and 7 OSAS patients were analyzed. Significant differences are found in the formant frequencies of both groups. The results are discussed from the view point of patho-physiological aspect.
Content may be subject to copyright.
Mustafa Çavuşoğlu1, Mustafa Kamaşak2, Tolga Çiloğlu3,Yeşim Serinağaoğlu3,Osman Eroğul4
1 Max Planck Instıtute for Biological Cybernetics, High Field MR Center, 72076, Tuebingen,Germany
2 İstanbul Technical University, Computer Engineering Department, 34469, İstanbul,Turkey
3 Middle East Technical University, Electrical and Electronics Engineering Department, 06530, Ankara
4 Gülhane Military Medical Hospital, Biomedical and Clinical Engineering Center 06018, Ankara
In recent years, several studies have shown the
relationship between snoring and obstructive sleep apnea
syndrome (OSAS). Besides time domain analysis of
snoring signal, the spectral features and shapes of snores
can be used to discriminate simple snorers and OSAS
patients. In this study, we propose a method to classify
simple snorers and OSAS patients based on spectral
envelope estimation of snoring signals. The formant
frequencies and corresponding bandwidths are computed
for both group, and the variation and consistency of the
formant distributions are computed. A total of 1400
snoring episodes from 7 simple snorer and 7 OSAS
patients were analyzed. Significant differences are found
in the formant frequencies of both groups. The results are
discussed from the view point of patho-physiological
Spectral analysis, formant frequency, OSAS, snore
1. Introduction
Snores are breathing sounds produced during the sleep.
Breathing triggers mechanical oscillations of the tissues
such as soft palate or tongue around the constriction due
to the relaxation of the tissues in the upper airway. The
American Sleep Disorders Association (ASDA) defined
snoring as “Loud upper airways breathing, without apnea
or hypoventilation, caused by vibrations of the pharyngeal
tissues” [1]. In the last 15 years, snoring problem has
entered the realm of clinical medicine. It is a prevalent
symptom, and about 50% of the adult population snores
frequently [2, 3]. It has been reported as a risk factor for
the development of diseases such as ischemic brain
infraction, systemic arterial hypertension, and coronary
artery disease and sleep disturbance [4]. Sleep apnea is
the cessation of airflow to the lungs during sleep for 10
sec. or more [5, 6]. It may result from either upper airway
collapse (obstructive sleep apnea) or lack of neural input
from the central nervous system to the diaphragm (central
sleep apnea) [6]. OSA is the more common form of sleep
apnea. Common symptoms of OSA are fatigue, daytime
sleepiness, coroner problems, and systemic hypertension
[5, 6]. It is usually associated with loud, heavy snoring
[6]. In recent years, several studies have shown the
relationship between snoring and obstructive sleep apnea
syndrome (OSAS). These studies have shown that the
frequency domain analyses of snores can be used to
discriminate simple snorers and OSAS patients [7].
Analyzing snoring sounds is very useful and promising
tool in sleep related breathing disorders. Time domain
parameters of snoring sounds such as the snore rate
(number of snoring episodes per hour), intensity of the
snoring and the regularity of the snoring episodes showed
that simple snorers and OSAS patients can be classified
by using snoring sound analysis instead of
polysomnographic studies which requires spending the
whole night in the hospital.
Hill et. al. [8] proposed the use of acoustic crest factor of
snoring episodes (the ratio of peak amplitude to root mean
square value) to distinguish palatal from nonpalatal snore.
Abeyratne et. al. [9] proposed the measure “intra-snore-
pitch-jump” to diagnose OSAS. Sola-Soler et. al. [10]
studied the differences in spectral envelopes of simple and
OSAS snores and suggested the standard deviation of
formant frequencies as a criterion for distinction among
simple snores and OSAS snores. Miyazaki et. al. [11]
found significant differences in fundamental frequency
depending on the snore production site. Spence et. al.
found a significant correlation between median frequency
of snore and apnea/hypopnea index (AHI). Cavusoglu et.
al. [12] developed an interface system based on time
domain characteristics to differentiate the simple snorers
and OSAS patients as well as to determine the efficiency
of applied snoring treatment objectively.
Most of the studies are focused on the fundamental
frequency of the snores but spectral envelope estimation
of the snoring signals also contains information about the
pressure pulses and anatomical filtering involved in snore
production [10]. Thus, the spectral analysis values, the
“formant type” structure and the shape of the spectrum
help to distinguish simple snorers and OSAS patients
[13]. Polysomnography (PSG), performed over a full
night sleep, is presently the gold standard for diagnosis of
sleep apnea [14, 15]. Since PSG analysis requires
spending the night at a hospital with many measurement
electreodes connected to the body, it is time consuming
and uncomfortable for the patient. The motivation of this
study was to use spectral envelope analysis of snoring
signals to differentiate the simple snorers and OSAS
patients as an alternative to polysomnography. This
analysis may also provide information in localization of
the snoring site such as palate or tongue.
Our approach in modeling snoring sounds is motivated by
the well-known source-filter modeling framework of
speech sounds. According to the source filter-model,
speech is produced by the excitation of the vocal tract (the
acoustic filter) by an acoustic pressure wave due to the
motion of vocal folds or due to a constriction created by
the speech articulators. The resulting excitation signal has
a mixed periodic and noise content in varying proportions
for different phonemes. Source-filter model is also very
suitable for modeling snoring sounds. The mechanism of
snoring is similar to that of voiced speech. The main
difference arises at the source part. The vibration of vocal
folds is replaced by the vibration of relaxed pharyngeal
tissues and tongue. Snorer’s relaxed tissues tend to come
into contact when there is no airflow for example, around
the intake-outlet turnover instants of breathing. Following
the start of an intake or outlet, the pressure difference
across the contact causes the separation of contacting
tissues. Once the tissues are separated, the pressure
difference drops and the tissues come into contact again
and the cycle continues in this manner. The location of
the contact in the pharynx, the type of contacting tissues,
the physical properties of the tissues depending on age
and/or fat content affects the dynamics of this vibration.
The excitation of the vocal and nasal tracts as a
consequence of this vibration yields the lip and nose
radiation of the snoring sound.
The formant frequencies are the resonant frequencies of
the vocal tract filter its range depends on the shape of the
resonant cavities. Since the source of the snoring sound
differs in simple snorers and OSAS patients, it motivates
to the investigation of how the formant frequencies and
their bandwidths changes in simple snorers and OSAS
2. Materials and Methods
2.1 Recording Set up and Database
A Sennhiser ME 64 condenser microphone with a 40–
20000 Hz ± 2.5 dB frequency response was used for
recording sounds. This microphone has a cardioid pattern
which helps to suppress some of the echoes from the
environment. It was placed 15 cm over the patient’s head
during sleep. The signal was fed via a BNC cable to the
Edirol UA-1000 model multi-channel data acquisition
system connected to a personal computer via universal
serial bus. The computer was placed outside the sleeping
room not to disturb the sleeping patients. The acquired
signal was digitized at a sampling frequency of 16 KHz
with 16 bit resolution.
A total of 1400 snoring episodes were analyzed obtained
from 7 simple snorers and 7 OSAS patients equally.
Snoring episodes were selected by an automatic detector
[15] and validated by a medical doctor. Apnea/Hyponea
Index (AHI) of simple snorers and OSAS patients are
4.29 and 39.21 respectively.
2.2 The Formant Type Spectrum Structure of
Snoring Sounds
Unlike the highly non-stationary character of speech
because of the continuous motion of the articulators (jaw,
lips, tongue, and velum) along the vocal tract, snoring
signal bears non-stationarity at a much lower level.
Articulators keep their positions unchanged most of the
time during snoring. Spectral envelope estimations of
snoring signals are shown in Figure 1. The spurious peaks
with very low amplitude are rejected by 3dB threshold.
Inspecting the snoring intervals clearly indicates the
highly stationary behavior of the signal. This is much like
the production of sustained vowels.
Figure 1 shows that the greater part of the energy content
is below 2500 Hz and the main components lie in the low
frequency range, at about 150 Hz. The envelope peaks
also at the frequencies about 500, 1800 Hz, and 2200 Hz.
The spectrum shows a fundamental frequency and a
“formants type” structure.
00.5 11.5 22.5 33.5 44.5 5
frequency (kHz)
Power (db)
Figure 1. Spectral envelope estimations of snoring signals
In the context of speech processing, formants are the
frequencies around which the peaks of the spectral
envelope are observed. They are the resonant frequencies
of the vocal tract filter. Each formant is characterized by
frequency, bandwidth and amplitude level, and its range
depends on the shape of the resonant cavities. The
different conditions in which the subjects and the patients
who snore can affect the formants range in the frequency
spectrum. The formants of a snoring signal including 5
snoring episodes is shown in Figure 2.
Figure 2. The formants of a snoring signal including 5
snoring episodes
Automatic spectral envelope estimation of the snores in
the database shows that formants stay almost unchanged
within and across the snoring intervals and the formant
frequencies of snoring episodes have almost common
formants in some frequency range.
2.3 Formant Frequency Analysis
We compute the formant frequencies based on linear
prediction coefficients (LPC) of the snoring episodes.
These coefficients are computed using auto-correlation
method with Levinson-Durbin recursions.
)()()( (1)
In equation (1), represents the snoring signal and p
is the degree of linear prediction. In addition, and
corresponds to the linear prediction coefficients
and prediction error respectively. We can express the
signal in frequency domain as follows:
)()()( (2)
we can modify (2) in order to obtain the transfer function
that is expressed by (3).
zE zX
zH p
The sound signal is generated by stimulating
the , location of the production (uvula, soft palate,
etc), with .The transfer function, is
determined by minimizing the linear prediction
(zE )(zH
The formant frequencies are computed by finding the
roots of the linear prediction polynomial obtained from
LPC analysis. The prediction polynomial can be
expressed as follows:
kkkk zczczazA
1*11 )1)(1(1)( (4)
where p is the order of the LPC filter. Resonance
frequencies are found using only the complex roots of
prediction polynomial. The complex root of the
is expressed in equation (5).
kk ecc = (5)
The formant frequency and the corresponding
bandwidth can be written as:
= (6)
= (7)
where is the sampling period. In this study, the order of
the linear prediction filter is selected as 10.
3. Results
We investigate the inter- and intra-patient variation of
formant frequencies and their bandwidths for simple
snorers and OSAS patients. To visualize these variations,
we have randomly selected 10 snoring episodes from each
of the 7 simple snorers and 7 OSAS patients and compute
the formant frequencies. Figure 3 and Figure 4 shows the
inter- and intra- patient variation of formant frequencies
in simple snorers and OSAS patients respectively.
Average formant frequencies and corresponding
bandwidths are calculated for the snoring episodes in the
database. Table 1 and Table 2 show the average values of
formant frequencies and corresponding bandwidths
respectively for simple snorers and for OSAS patients.
Figure 3. Inter- and intra- patient variation of formant
frequencies in simple snorers
Figure 4. Inter- and intra- patient variation of formant
frequencies in OSAS patients
Table 1
Average values of formant frequencies for Simple snorers
and OSAS patients
Subject F1 F2 F3 F4
snorer 512.8 1678.2 4326.5 6728.9
patient 112.4 1386.6 4203.8 6404.2
Table 2
Average values of formant bandwidths for Simple snorers
and OSAS patients
Subject BW1 BW2 BW3 BW4
snorer 9.4 18.6 19.8 28.6
patient 27.5 64.3 44.6 31.9
4. Discussion and Conclusion
The production mechanism of snoring sound is similar to
that of voiced speech; hence we can apply the well-known
source-filter model to analyze the snoring sounds.
Spectral envelope analysis of snoring signals provides an
efficient way in simple snorer/OSAS classification which
is very important both in the diagnostic and treatment or
uvulo-palatopharyngoplasty (UPPP) type surgery process
of patients. Filtering characteristics of upper airway and
the location of pressure production are strongly related to
the spectral envelope. For example, the shape of resonant
cavities in the upper airway changes the range and value
of formant frequencies [10].
In simple snorers (Figure 3), while the formant
frequencies are consistent in different snoring episodes of
same individual, there may be some shifting in different
patients. The magnitude of these shifting is less in the first
formant frequency than those of other three formant
frequencies. The intra-patient consistency of formant
frequencies is highest in the first formant and lowest in
the second formant. For each patient, the formant
frequencies may vary. However, from figure 2, it can be
seen that the formant variations are correlated among the
In OSAS patients (Figure 4), there is no considerable
coincidence in formant frequency locations both inter-
and intra patient snoring episodes. The significantly
lower variability of formant distributions in simple
snorers than in OSAS patients can be explained as
i) The source of the snoring sound (vibrating tissue in
the upper airway) is in a multi-segmental structure in
OSAS patients.
ii) In OSAS patients, upper airway resistance is
relatively lower compared to simple snorers. The
reduction in the upper airway resistance is related to
the site of snoring.
In this study, we propose a method to discriminate the
simple snorers and OSAS patients based on spectral
envelope estimation of snoring signals. The experimental
results are promising and show that the approach given
here can be used not only for source localization of
snoring but also for detection of the obstructive sleep
[1] Thorpy.MJ “The international classification of sleep
disorders: diagnostic and coding manual”. Lawrence KS,
ed. Allen Press Inc., USA 1990; 195–197.
[2] Lugaresi.E, Cirignotta.F, Coccagna.C, PianaC., “Some
epidemio-logical data on snoring and cardiorespiratory
disturbances” Sleep 1980; 3:221-224
[3] Norton.PG, Dunn .EV “Snoring as a risk factor for
disease: an epidemiological survey.” Br. Med J 1985;
[4] K. Wilson, R.A Stoohs, T.F. Mulrooney, L.J. Johnson,
C. Guilleminault, Z. Huang, “The snoring
spectrum:acoustic assesment of snoring sound intensity in
1,139 individuals undergoing polysomnography” Chest
1999, vol 115:3 762-70,
[5] J.Lucas, J. Golish, G. Sleeper, and J.A.O’Ryan,
Home Respiratory Care”. Englewood Cliffs, NJ:
Prentice Hall, 1988, ch. 6, 132-136.
[6] R.L. Wilkin, “Clinical Assessment in Respiratory
Care.” St Louis, MO: Mosby, 1985,268-272.
[7] J.A. Fiz, J.Abad, R. Jane, M. Riera, M.A. Mananas, P.
Caminal, D. Rodenstein, J. Morera, “Acoustic Analayses
of snoring sound in patients with simple snoring and
OSA” European Respiration Journal, vol 9, 2365-2370,
[8] Hill P.D., Lee B.V, Osborne J.E., Osman E.Z.,
Palatal snoring identified acoustic crest factor
analysis”, Physiol. Meas. 20 167-174, 1999
[9] Abeyratne U R, Wakwella A S and Hukins C 2005
Pitch jump probability measueres for the analysis of
snoring sounds in apnea” Physiol. Meas. 26 779-98
[10] J. Sola-Soler, R. Jane, J.A. Fiz, J. Morera, “Spectral
Envelope Analysis in Snoring Signals from Simple
Snorers and patients with Obstructive Sleep Apnea”
Proceeding of the 25th
International Conference of the
IEEE EMBS, Cancun, Mexico, September 17-21 2003
[11] Miyazaki S., Itasaka Y., Ishıkaea K., Togawa K,
Acoustic analysis of snoring and the site of airway
obstruction in sleep related respiratory disorders”, Acta
Otolaryngol Suppl(Stockh), 1998, 537, 47-51
[12] Cavusoglu M, Kamasak M, Erogul O, Ciloglu T,
Akcam T and Serinagaoglu Y “SASA:A Software
System for Sleep and Snore Analysis” Proc. of Int. Symp.
on Health Informatics and informatics, Antalya2007
[13] Dalmasso F and Prota R 1996 “Snoring: analysis,
measurement, clinical implications and applications”
Eur. Respir. J.9 146–59
[14]. C. Guilleminault, R. Stoohs, A. Clerk, M. Cetel, and
P. Maistros, A cause of excessive daytime sleepiness-The
The upper airway resistance syndrome, Chest, vol.
[15]. J.H. Peter, T. Podszus, P. von Wichert, “Sleep
Related Disorders and Internal Diseases”, New York:
Springer-Verlag 1987, pp, 101-107
ResearchGate has not been able to resolve any citations for this publication.
Full-text available
CLINICAL ASSESSMENT OF RESPIRATORY DISTRESS IN A NEONATE The neonates commonly suffer from respiratory diseases. Assessment of lung function is essential for appropriate and timely respiratory intervention. As on date, specific lung functions remain highly experimental tool and do not have common practical applicability. Therefore, common clinical signs could be used as indicator of lung functions and thus can effectively guide management. The following variables could be used bedside, for assessment of lung functions. 1. Respiratory rate and depth: The respiratory rate of 60 per minute and or increased depth signify respiratory disease. Their variation may suggest specific diseases. Decreased rate indicates decreased alveolar ventilation. Increased rates may be associated with increased dead space ventilation. The depth decreases in cases of rigid lungs because of poor lung compliance, as in hyaline membrane disease. On contrary, in obstructive lung disease with increased airway resistance, there is increased depth of respiration. 2. Retractions: The normal respiratory movements occur as a consequence of action of respiratory muscles on stable chest wall causing changes in intrapleural pressures and in lung volumes. In diseases, where lungs have poor compliance and are stiff, the effect of changes in the intrapleural pressures manifest more on chest wall and hence retractions appear. More stiffer the lungs more are the retractions. The retraction is suprasternal or infrasternal or intercostal. Suprasternal retractions more often suggests upper airway obstruction and may be a pointer toward upper airway anomaly in neonates. Intercostal retraction suggests alveolar involvement. 3. Nasal flaring: It occurs due to contraction of alae nasi muscle with resultant increase in nostril size and decrease in airway resistance. It is important to note that neonates are preferential nose breathers and thus airway resistance of nostril plays a very crucial part in lung functions. Nasal flaring is modulated by chemo-and mechanoreceptors. 4. Grunting: The normal vocal cord position of abduction in inspiration and adduction in expiration, do not produce any noise. In disease states, vocal cord remains partially adducted state with resultant production of grunting. Intensity and persistence of grunting are proportional to severity of disease. This signifies the patient’s effort to maintain functional residual capacity. The patient exhales against a partially closed glottis thereby attempting to keep the alveoli open. This again suggests presence of atelectasis. 5. Cyanosis: Presence of cyanosis indicates significant right to left shunt resulting from widespread atelectasis. Objectivity for assessment and monitoring for respiratory distress may be achieved by using Silverman-Anderson scoring or Downe’s scoring system. In Silverman-Anderson score, inspection or auscultation of the upper and lower chest and nares are scored on a scale of 0, 1 or 2 using this system are: Silverman-Anderson scoring system Features Score 0 Score 1 Score 2 Chest Movement Equal Respiratory Lag See-saw Respiration Intercostal retraction None Minimal Marked Xiphoid retractions None Minimal Marked Nasal flaring None Minimal Marked Expiratory grunt None Audible with stethoscope Audible without stethoscope a. Chest movement: Synchronized vs minimal lag or sinking of the upper chest as the abdomen rises. In the most extreme instances, a seesaw-like movement of the chest and abdomen is observed and would be given a score of 2. b. Intercostal retractions: Retraction between the ribs is rated as none,minimal or marked. This indicates loss of functional residual capacity. Xiphoid retractions: Similarly, retraction below the xiphoid process are rated as none, minimal or marked c. Nasal flaring: Normally, there should be no nasal flaring. Minimal flaring is scored 1 and marked flaring is scored 2. d. Expiratory grunting: Grunting that is audible with a stethoscope is scored 1, and grunting that is audible without using a stethoscope is scored 2. The higher the score, the more severe the respiratory distress. A score greater than 7 indicates that the baby is in respiratory failure. Down’s scoring system Features 0 1 2 Cyanosis None In room air In 40% FiO2 Retractions None Mild Severe Grunting None Audible with stethoscope Audible without stethoscope Air entry Clear Decreased or delayed Barely audible Respiratory rate Under 60 60-80 Over 80 or apnea Score: > 4 = Clinical respiratory distress; monitor arterial blood gases > 8 = Impending respiratory failures This scoring system is very practical and easy to use with good sensitivity and specificity in preterm babies. References: 1) Dr. Sanwar Agrawal, “Paediatrics For You” Raipur (M.P.) 1996;2:97-98. 2) Bhakoo ON. Assisted ventilation in neonates: The Indian Perspective. Indian Pediatr 1995; 32: 1261-1264. 3) Engle WA. Surfactant Replacement therapy for Respiratory Distress in Preterm and Term Neonate. Paediatrics 2008; 121: 419-432. Ms. Rameshwari Solanki Lecturer MTIN, Changa.
The normal function of the organism in various stages of activity can be seen as a process of mutual interaction of different regulation mechanisms building up the behaviour of the organism in changing situations and/or ages and/or levels of health or disease. Insights into these very complex relations which steer the action of the organism are particularly important for the study of internal medi­ cine. They can be seen as directly related to the understanding of pathological conditions. Recently, the medical community has focused its interest on the physiology and pathophysiology of events which happen during sleep. Although some information on pathological regulation during sleep was collected in the 1930s, the modern technology of registration methods is required to analyse the pheno­ mena of sleep-related physiological and pathophysiological patterns. It is intere­ sting that the modern research in this field developed from neuropsychiatry and is now expanding into other fields of medicine, although some problems, for instance the Pickwickian syndrome, were also for a long time considered part of the field of internal medicine. It becomes clearer that sleep is not only a neuro­ psychiatric phenomenon, but also has profound consequences for other physio­ logical circuits, perhaps even an important role in pathogenesis. Sleep also has profound consequences for internal diseases. This was shown very clearly recently by many groups, although their data have not yet received the attention which they deserve.
In a study conducted in four family practice units in Toronto, Canada, 2001 subjects reported on snoring and medical conditions in members of their households. For spouses the prevalence of snoring increased with age up to the seventh decade, with a higher prevalence of nearly 85% in husbands. For 11 medical problems an association existed between snoring, its frequency, and the presence of the condition. This association continued when the data were corrected for sex, age, and marital state. For hypertension both men and women who snored between the fifth and 10th decades had a twofold increase over non-snorers. The prevalence of heart disease and other conditions, except for diabetes and asthma, also increased in snorers in this age group. When corrected for smoking and obesity the association between snoring, hypertension, and heart disease persisted. These findings extend those of Lugaresi et al, and if they could be confirmed snoring as a risk factor for conditions other than sleep apnoea and sleep disorders might be considered. Methods of alleviating the acoustic annoyance of snoring may also provide direct medical benefits.
Subjects with isolated complaints of chronic daytime sleepiness are usually classified as "idiopathic hypersomniacs" and treated symptomatically. A group of these subjects was investigated during nocturnal sleep and daytime naps. In a subgroup of them, sleep was fragmented by very short alpha EEG arousals throughout the sleeping period. These short arousals are usually ignored in sleep analyses, but their impact is significant (in the 15 subjects identified with the syndrome, the mean sleep latency in multiple sleep latency tests was 5.1 +/- 1 min). These arousals are directly related to an abnormal increase in respiratory efforts during sleep (the mean peak inspiratory esophageal pressure measured in our subjects in the respiratory cycle just preceding a transient arousal was -33 +/- 7 cm H2O). Typically, an arousal occurs within one to three breaths of flow limitation associated with abrupt but limited reduction in tidal volume (ie, abnormal increase in upper airway resistance during sleep). The arousal restores normal breathing. Snoring was noted in association with these transient arousals in 10 of the 15 subjects; however, snoring was neither sufficient nor necessary for the identification of the clinical syndrome. Both sexes were equally represented in the affected group. All studied subjects had upper airway anatomy that was mildly abnormal. Nasal continuous positive airway pressure, used as an experimental tool, eliminated the daytime sleepiness (multiple sleep latency mean score = 13.5 min), the transient arousals (mean alpha EEG arousal index decreased from 31.3 +/- 12.4 to 8 +/- 2 per hour of sleep), and the abnormal upper airway resistance. Chronic daytime sleepiness is a major cause of social, economic, and medical impairment. Recognition of this syndrome and its cause is important, as specific treatments can be developed to eliminate the problem.
Snoring was described in literature even before medicine. Common definitions do not consider acoustic measurements of snoring. In this paper we discuss the main pathophysiological aspects of snoring and the snoring-sleep relationship as the generating mechanisms. Snoring can be analysed and measured by the following methods: 1) Leq-Equivalent Continuous Sound Level, which only quantifies noisiness, annoyance, and damage to the partner's and snorer's hearing; 2) Power Spectrum, with frequency values, formantic structure data and typical shape, which can help to distinguish simple snoring from heavy snoring with obstructive sleep apnoea syndrome (OSAS); 3) Linear Prediction Code (LPC) method, which can define the cross-sectional area (CSA) of the upper airways and which locates sites of obstruction. Simulated snoring analysis with LPC and with simultaneous fluoroscopy permits the definition of CSA and the identification of three snoring patterns: nasal, oral and oronasal. Snoring is an important sign of sleep-related breathing disorders (SRBD), of the upper airway resistance syndrome (UARS), and of the OSAS. Snoring is a symptom of nasal obstruction and is associated with cardiovascular diseases and nocturnal asthma as a trigger or causative factor; however, its acoustic features in these disorders are not well-defined. Home monitoring of snoring is very useful for epidemiology and is mandatory, together with heart rate and arterial oxygen saturation (Sa,O2), to screen SRBD.
Seventy-five adult patients with sleep related respiratory disorders were examined by polysomnography with simultaneous recordings of the intraluminal pressure of the upper airway and snoring sound. Obstructed sites in the upper airway during sleep were determined by comparing the amplitude of respiratory fluctuation of the pressures in the epipharynx, mesopharynx, hypopharynx and esophagus. A definite correlation existed between the intensity of snoring sound and the amplitude of respiratory fluctuation of the intraesophageal pressure. Based on the results of the intraluminal pressure partitioning, the subjects were divided into the soft palate type (28), the tonsil/tongue base type (14), the combined type (27) and the larynx type (6). The average value of fundamental frequency (ff) was 102.8+/-34.9 Hz in the soft palate type, 331.7+/-144.8 Hz in the tonsil tongue base type, 115.7+/-58.9 Hz in the combined type and around 250 Hz in the larynx type.
To quantify the snoring sound intensity levels generated by individuals during polysomnographic testing and to examine the relationships between acoustic, polysomnographic, and clinical variables. The prospective acquisition of acoustic and polysomnographic data with a retrospective medical chart review. A sleep laboratory at a primary care hospital. All 1,139 of the patients referred to the sleep laboratory for polysomnographic testing from 1980 to 1994. The acoustic measurement of snoring sound intensity during sleep concurrent with polysomnographic testing. Four decibel levels were derived from snoring sound intensity recordings. L1, L5, and L10 are measures of the sound pressure measurement in decibels employing the A-weighting network that yields the response of the human ear exceeded, respectively, for 1, 5, and 10% of the test period. The Leq is a measure of the A-weighted average intensity of a fluctuating acoustic signal over the total test period. L10 levels above 55 dBA were exceeded by 12.3% of the patients. The average levels of snoring sound intensity were significantly higher for men than for women. The levels of snoring sound intensity were associated significantly with the following: polysomnographic testing results, including the respiratory disturbance index (RDI), sleep latency, and the percentage of slow-wave sleep; demographic factors, including gender and body mass; and clinical factors, including snoring history, hypersomnolence, and breathing stoppage. Men with a body mass index of > 30 and an average snoring sound intensity of > 38 dBA were 4.1 times more likely to have an RDI of > 10. Snoring sound intensity levels are related to a number of demographic, clinical, and polysomnographic test results. Snoring sound intensity is closely related to apnea/hypopnea during sleep. The noise generated by snoring can disturb or disrupt a snorer's sleep, as well as the sleep of a bed partner.
The differentiation of palatal from non-palatal snoring is very important for ENT surgeons trying to determine whether palatal surgery would be curative. At present sleep nasendoscopy is the accepted method. Palatal vibration produces marked modulation of sound loudness at low frequency (below ~100 Hz). We calculate a crest factor for the sound waveform (ratio of peak to root mean square (rms) value in any given epoch), as a measure of the degree of modulation. Free-field snore sounds were recorded from 11 supine adult patients under intravenous sedation (midazolam), using a digital tape recorder. Recordings were transferred to a PC (sampling frequency 11 kHz), and analysed using code written by us. Direct visual confirmation of the site of snoring was gained from simultaneous sleep nasendoscopy, taken as the gold standard. In six patients the dominant site was the soft palate. The non-palatal group (five patients) comprised one epiglottic, two hypopharyngeal and two tongue base snorers. The crest factor was found to be significantly higher for palatal snorers (p<0.01, Student-t or Mann-Whitney tests). Furthermore, palatal could be distinguished from non-palatal snorers on the basis of crest factor alone in all 11 cases, making this a promising non-invasive diagnostic technique.