Acoustic analysis of snoring sound in patients with simple snoring and obstructive sleep apnoea

Eu Respi J, 1996, 9, 2365–2370 DOI: 10.1183/09031936.96.09112365 P in ed in UK - all igh s ese ed Copy igh ERS Jou nals L d 1996 Eu opean Respi a o y Jou nal ISSN 0903 - 1936 AAccoouuss iicc aannaallyyssiiss oo ssnnoo iinngg ssoouunndd iinn ppaa iieenn ss wwii hh ssiimmppllee ssnnoo -- iinngg aanndd oobbss uucc ii ee sslleeeepp aappnnooeeaa J.A. Fiz*, J. Abad*, R. Jané**, M. Rie a*, M.A. Mañanas**, P. Caminal**, D. Rodens ein+, J. Mo e a* Acous ic analysis o sno ing sound in pa ien s wi h simple sno ing and obs uc i e sleep apnoea. J.A. Fiz, J. Abad, R. Jané, M. Rie a, M.A. Mañanas, P. Caminal, D. Rodens ein, J. Mo e a. ©ERS Jou nals L d 1996. ABSTRACT: Sno ing, a symp om which may indica e he p esence o he obs uc- i e sleep apnoea synd ome (OSA), is also common in he gene al popula ion. Recen s udies ha e sugges ed ha he acous ic cha ac e is ics o sno ing sound may di e be ween simple sno e s and OSA pa ien s. We ha e s udied a small numbe o pa ien s wi h simple sno ing and OSA, analysing he acous ic cha ac- e is ics o he sno ing sound. Se en een male pa ien s, 10 wi h OSA (apnoea/hypopnoea index (AHI) 26.2 e en s·h-1) and se en simple sno e s (AHI 3.8 e en s·h-1), we e s udied. Full nigh polysomnog aphy was pe o med and he sno ing sound powe spec um was ana- lysed. Spec al analysis o sno ing sound showed he exis ence o wo di e en pa - e ns. The i s pa e n was cha ac e ized by he p esence o a undamen al e- quency and se e al ha monics. The second pa e n was cha ac e ized by a low equency peak wi h he sound ene gy sca e ed on a na owe band o equen- cies, bu wi hou clea ly iden i ied ha monics. The se en simple sno e s and wo o he 10 pa ien s wi h OSA (AIH 13 and 14 e en s·h-1, espec i ely) showed he i s pa e n. The es o he OSA pa ien s showed he second pa e n. The peak equency o sno ing was signi ican ly lowe in OSA pa ien s, wi h all bu one OSA pa ien and only one simple sno e showing a peak equency below 150 Hz A signi ican nega i e co ela ion was ound be ween AHI and peak and mean equencies o he sno ing powe spec um (p<0.0016 and p<0.0089, espec i ely). In conclusion, his s udy demons a es signi ican di e ences in he sound powe spec um o sno ing sound be ween subjec s wi h simple sno ing and obs uc i e sleep apnoea pa ien s. Eu Respi J., 1996, 9, 2365–2370. *Se ei de Pneumologia, Hospi al Uni e si a io Ge mans T ias i Pujol, Badalona, Spain. **Ins i u o de Cibe né ica (UPC), Ba celona, Spain. +Cliniques Uni e si ai es Sain Luc, Uni e si é Ca holique de Lou ain, B ussels, Belgium. Co espondence: J.A. Fiz Se ei de Pneumologia Hospi al Ge mans T ias i Pujol Ca e e a del Canye s/n 08916 Badalona Ba celona Spain Keywo ds: Acous ic analysis obs uc i e sleep apnoea sno e Recei ed: Ap il 10 1995 Accep ed a e e ision June 20 1996 This wo k was suppo ed by g an 94/0625 FIS (Spain). In he las 15 y s, sno ing has en e ed he ealm o clinical medicine. Sno ing may dis u b he social and amily li e o he sno e , who is unawa e ha he sno es. Sno ing is a p e alen symp om, and abou 50% o he adul popula ion sno es equen ly [1, 2]. One o i e pe cen o he adul male popula ion su e om he sleep apnoea synd ome, in which sno ing is a dominan symp om. Se e al ea men s exis o simple sno ing and o obs uc i e sleep apnoea (OSA), many o hem mu ila ing o in asi e. Recen s udies ha e sugges ed ha he acous ic cha ac e is ics o sno ing may di e be ween simple sno e s and pa ien s wi h OSA. Heal hy simple sno e s, wi hou apnoea episodes, showed a powe spec um o sno ing signal wi h a ha monic s uc- u e and a undamen al equency ha anged 110–190 Hz [3]. In sno ing pa ien s wi h OSA, equency com- ponen s highe han 800 Hz we e ound [4]. We ha e s udied a small numbe o pa ien s wi h sim- ple sno ing o OSA, and analysed he acous ic cha ac- e is ics o he sno ing sound as well as he pa e n o espi a o y dis u bances du ing sleep. Ma e ial and me hods Se en een li elong nonsmoking male pa ien s om he sleep diso de s clinic o he Ge mans T ias i Pujol hospi al in Badalona we e s udied. Ten o he subjec s we e diagnosed as ha ing OSA and se en as ha ing simple sno ing. Lung unc ion es s we e pe o med using a spi o- me e (PFT Ho izon Sys em) [5]. Re e ence alues we e hose o ROCA e al. [6]. A e ial blood gas alues we e measu ed using a blood gas analyse (Radiome e ABL). Full nigh polysomnog aphy was pe o med acco d- ing o s anda d me hods [7]. Sleep analysis was pe - o med by isual inspec ion using RECHTSCHAFFEN and KALES [7] pape sco ing. Elec oencephalog am (EEG), elec o-oculog am (EOG) and chin elec omyog am (EMG) we e ob ained om su ace elec odes. Tho acic and abdominal espi a o y mo emen s, as well as hei sum, we e ob ained using a induc i e ple hysmog aph (Respi- ace NIMS, Miami, USA). Calib a ion was pe o med by means o a d y spi ome e (S3371; Senso Medics, J.A. FIZ ET AL. 2366 Table 1. – An h opome ic and espi a o y unc ion da a o subjec s wi h simple sno ing and pa ien s wi h obs uc- i e sleep apnoea Simple sno ing OSA Age y s 46±13 51±6 Heigh cm 171±9 166±7 Weigh kg 87±22 91±21 BMI kg·m-2 29.7±7.2 32.9±7.6 FVC L 4.4±1.1 3.5±0.6 % p ed 90±18 83±13 FEV1L 3.5±1.1 2.8±0.7 % p ed 93±20 85±17 FEV1/FVC % 81±3 78±8 Pa,O2kPa 12.2±0.8 10.6±2.1 mmHg 91.3±5.5 79.7±16.0 Pa,CO2kPa 5.0±0.3 5.4±1.0 mmHg 37.2±2.6 40.2±7.8 Sa,O2% 96±1 94±6 pH 7.37±0.0 7.39±0.0 Values a e p esen ed as mean±SD. FVC: o ced i al capaci y; BMI: body mass index; % p ed: pe cen age o p edic ed alue; FEV1: o ced expi a o y olume in one second; Pa,O2: a e ial oxygen ension; Pa,CO2: a e ial ca bon dioxide ension; Sa,O2: a e ial oxygen sa u a ion. Table 2. – Sleep pa ame e s o subjec s wi h simple sno ing and pa ien s wi h obs uc i e sleep apnoea (OSA) Simple sno ing OSA TST min 268±66 315±68 Awake % o TST 23±13 10±8* S age 1 % o TST 41±18 28±17 S age 2 % o TST 28±18 51±16 S age 3&4 % o TST 1.3±3.3 0.3±0.9 REM % o TST 4.7±4.6 10.8±11.7 Lowe Sa,O2% 92±4 80±15 Mean Sa,O2% 95±2 90±6 AHI e en s·h-1 3.8±1.7 26.2±20.1 Maximum du a ion o e en s s 26.7±16.5 42.7±27.3 Mean du a ion o e en s s 17.7±3.7 20.5±3.9 Values a e p esen ed as mean±SD. TST: o al sleep ime; REM: apid eye mo emen (sleep); Sa,O2: a e ial oxygen sa - u a ion; AHI: apnoea/hypopnoea index *:p<0.05, o com- pa isons be ween g oups (Kolmogo o -Smi no nonpa ame ic es ). Anaheim, USA). A e ial oxygen sa u a ion (Sa,O2) was measu ed using a pulse oxime e (Oxy Shu le; Senso Medics, USA) wi h a inge p obe. O onasal low was assessed using he mis o s. All he signals we e eco d- ed on pape using a Senso medics eco de . In addi ion o he abo e, a minia u e mic ophone (MKE 3010; SennHeise , Ge many) was posi ioned upon he neck, 1 cm la e al o he median line a he le el o he c icoid ca ilage. The mic ophone was enclosed in a plas ic hemisphe e, 3 cm in diame e and lea ing a 1 cm dis ance be ween he mic ophone and he skin, a oiding di ec con ac wi h skin su ace. The hemisphe e was a a- ched o he skin using adhesi e ape. The mic ophone had a la equency esponse be ween 40 Hz and 30 kHz ( ield ansmission ac o in open loop: 10 m ·Pa-1 ±2.5 dB). An apnoea was de ined as he absence o ai low du - ing ≥10 s. Apnoeas we e classi ied as obs uc i e o cen al acco ding o he pe sis ency o absence o es- pi a o y mo emen s, accompanied by a all in Sa,O2o ≥4%. Hypopnoea was de ined as a educ ion in idal olume o ≥50% o hose eco ded du ing he p eced- ing i e b ea hs o longe han 10 s, accompanied by a all in Sa,O2o ≥4%. OSA was de ined as he p es- ence o mo e han 10 apnoeas/hypopnoeas·h-1 o sleep, and he o al numbe o episodes o apnoea and hypop- nea pe hou o sleep ep esen ed he apnoea/hypop- noea index (AHI). The sound signal was il e ed wi h a band-pass il e be ween 10 Hz and 6 kHz (KH 39188). The sound sig- nal oge he wi h he low signal o he he mis o , was s o ed using a ideo ape casse e (Racal V S o e CH) a a speed o 7.5 in·s-1. Flow and sound signals we e p ocessed using a pe sonal compu e (PC Compaq desk 368/20e) wi h a sampling equency o 12 kHz. Analysis was limi ed o S age 1 and 2 non- apid eye mo emen (REM) sleep. Sno ing and low signals we e synch onized by means o a digi al clock a ailable on ideo ape eco de . Th ee episodes o h ee consecu i e b ea hs wi h sno ing we e andomly analysed a e iden i ica ion o 10 pe iods o sno ing. Analysis was pe o med using he a e age o he h ee episodes. In pa ien s wi h OSA, sno ing co - esponded o he 1s , 2nd and 3 d b ea h a e an apnoea. Sno ing was iden i ied by lis ening o he sound. Fas Fou ie T ans o m (FFT) was used o calcula e sound spec a o he h ee successi e sound inspi a o y episo- des o sno ing. FFT was applied on 1,024 samples. Welch pe iodog am was applied using a Hanning win- dow wi h an o e lap o 50%. The ollowing pa ame e s we e measu ed: maximal equency ( max), de ined as he uppe equency con- aining 90% o he o al powe o he spec um; peak equency ( peak) de ined as he equency wi h he max- imum powe ; and mean equency ( mean) de ined as he equency including hal he o al powe o he spec- um. Ha monics we e de ined as b oad equency bands in he spec um analysis [3, 4] wi h a undamen al e- quency. Fo each pa ame e , he a iabili y was es i- ma ed by calcula ing he coe icien o a ia ion (CoV) indi idually o each subjec o pa ien . Compa isons we e pe o med using he nonpa ame ic Kolmogo o - Smi no es . Spea man's nonpa ame ic co ela ion co- e icien was used o e i y he ela ionship be ween sleep pa ame e s wi h acous ic spec al sound pa ame- e s. Resul s we e conside ed signi ican i he p- alue was lowe ha 0.05. Resul s Table 1 p esen s he an h opome ic as well as he lung unc ion and blood gas alues o he pa ien s. Pa ien s wi h OSA we e olde , and had somewha lowe alues o o ced i al capaci y (FVC), o ced expi a- o y olume in one second (FEV1) and a e ial oxygen ension (Pa,O2) han pa ien s wi h simple sno ing. How- e e , he e we e no s a is ically signi ican di e ences be ween g oups. Polysomnog aphic da a a e ep esen ed in able 2. Sleep a chi ec u e was abno mal in bo h g oups, wi h an excess o S ages 1 and 2 non-REM sleep and e y ACOUSTIC ANALYSIS OF SNORING 2367 Fig. 1. – Analysis o sno ing signal om a simple sno e . a) Signal (a bi a y uni s (au)) o i s b ea h; b) powe spec um (au) o i s b ea h; c) signal (au) o second b ea h; d) powe spec um (au) o second b ea h. The e is a undamen al equency and clea ly iden i- ied ha monics. Fig. 2. – Analysis o sno ing signal om an obs uc i e sleep apnoea (OSA) pa ien . a) Signal (a bi a y uni s (au)) o i s pos apnoeic sno e; b) powe spec um (au) o i s pos apnoeic sno e; c) signal (au) o second pos apnoeic sno e; d) powe spec um o second pos apnoeic sno e. Maximum ene gy is a ound he undamen al equency; no e ha he e a e no inden i iable ha monics. Signal au -0.4 -0.2 0 0.2 0.4 a) 00.2 0.4 0.6 0.8 11.2 1.4 1.6 1.8 2.0 Time s 1.5 1.0 0.5 0 Powe spec um au 100 200 300 1000 900 400 500 600 700 800 0 F equency Hz 0.6 0.4 0.2 0 -0.2 -0.4 00.5 1.0 1.5 2.0 2.5 Time s Signal au b) c) 0.1 0.2 0 Powe spec um au 100 200 300 1000 900 400 500 600 700 800 0 0.5 0.4 0.3 d) F equency Hz 2 1 0 -1 -2 00.2 0.4 0.6 0.8 1.0 1.2 Signal au Time s 2.5 2.0 1.5 1.0 0.5 00 100 200 300 400 500 600 700 800 900 F equency Hz Powe spec um au 0 -1 -2 1 2 Signal au 00.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Time s 25 20 15 10 5 0 0 100 200 300 400 500 600 700 800 900 1000 F equency Hz a) b) c) d) 1000 Powe spec um au small amoun s o S ages 3 and 4 non-REM sleep. REM sleep was be e p ese ed. Spec al analysis o sno ing showed he exis ence o wo di e en pa e ns. The i s pa e n was cha ac e - ized by he p esence o a undamen al equency and se e al ha monics h ough a b oad equency band ( ig. 1). The second pa e n was cha ac e ized by he clea p edominance o sound ene gy sca e ed on a na owe band o equencies, bu wi hou clea ly iden i ied ha - monics ( ig. 2). The se en simple sno e s and wo o he 10 pa ien s wi h sleep apnoea showed he i s pa e n. The wo pa - ien s wi h he less se e e OSA showed he i s pa e n and had AHI o , espec i ely, 13 and 14 e en s·h-1 o sleep. The es o he pa ien s wi h OSA showed he second pa e n o sound spec um. Table 3 p esen s he a e age equency indices cal- cula ed o bo h g oups, as well as he mean o he indi- idual CoVs o each pa ame e . The e was a signi ican di e ence be ween pa ien s wi h simple sno ing and pa ien s wi h OSA in peak equency ( able 3). Using a h eshold o 150 Hz o he peak equency, all bu one o he pa ien s wi h OSA, bu only one subjec wi h sim- ple sno ing, showed alues below his h eshold. All o he o he indices showed a endency o lowe alues in pa ien s wi h OSA, bu he di e ences did no each s a is ical signi icance ( able 3). Signi ican nega i e co ela ions we e ound be ween he numbe o AHI·h-1 o sleep and he peak and mean equencies o he powe spec um ( ig. 3 and 4). Discussion In his s udy i was ound ha pa ien s wi h simple sno ing had a sno ing sound spec um cha ac e ized by a undamen al equency and he p esence o ha mon- ics. Sno ing analysis was pe o med on he i s b ea hs ollowing an apnoea in pa ien s wi h OSA. The pa e n dec ibed o simple sno e s was only a ely seen in pa - ien s wi h OSA, who ins ead seemed o ha e a spec- um cen ed a ound a undamen al equency wi hou ha monics. In addi ion, peak equency was lowe in pa ien s wi h sleep apnoea. A h eshold o 150 Hz bes iden i id pa ien s wi h sleep apnoea om simple sno - e s. Finally, a signi ican nega i e co ela ion was ound be ween he se e i y o he sleep apnoea synd ome, assessed by he AHI, and he sno ing sound spec um. Sno ing sound is p oduced by oscilla ions in he so pala e, pha yngeal walls and epiglo is [8–10]. Sno ing is always p eceded by low limi a ion [9]. Flow limi a- ion has been a ibu ed o he sleep- ela ed dec ease in pha yngeal muscle one. The pha ynx may be seen as a collapsible ube. A dec ease in he one o i s walls leads o a dec ease in i s c oss-sec ional a ea, wi h low limi a ion ollowed by high equency oscilla ions in he pha yngeal walls. In pa ien s wi h sleep apnoea, ade- qua e ai low is comp omised and is es o ed wi h he sno ing sound, only a e he a ousal ela ed eopening o he pha ynx. A ha ime, he e a e pa ial and apid closu es and openings o he pha yngeal lumen [11]. S udies on heo e ical models ha e ecen ly con i med he abo e hypo hesis on sno ing p oduc ion p oposed by LIISTRO e al. [11]. J.A. FIZ ET AL. 2368 Table 3. – Spec um equency pa ame e s o sno ing in subjec s wi h o wi hou obs uc i e sleep apnoea syn- d ome (OSAS) Sno e s CoV OSAS CoV %% peak Hz 264±107 14 157±136 14 mean Hz 325±58 15 223±147 16 max Hz 462±85 8 455±199 8 Values a e p esen ed as mean±SD o he ele an da a, and mean o he coe icien o a ia ion. peak: he equency wi h he highes powe ; mean: he equency wi h he mean spec- um powe ; max: he equency wi h 90% o o al spec um powe ; CoV: in asubjec coe icien o a ia ion. *: p<0.05, Kolmogo o -Smi no nonpa ame ic es . 600 500 400 300 200 100 0 010 20 30 40 50 60 70 80 AHI e en s·h-1 peak Hz Fig. 4. – Rela ionship be ween apnoea/hypopnoea index (AHI) and mean equency ( mean) o spec um in se en simple sno e s and 10 obs uc i e sleep apnoea (OSA) pa ien s. The e is a signi ican nega- i e co ela ion (Spea man ank o de co ela ion: =-0.61; p<0.0089). Fig. 3. – Rela ionship be ween apnoea/hypopnoea index (AHI) and peak equency ( peak) o spec um in se en simple sno e s and 10 obs uc i e sleep apnoea (OSA) pa ien s. The e is a nega i e ela- ionship. The e is a signi ican nega i e co ela ion (Spea man ank o de co ela ion: =-0.70; p<0.0016). AHI e en s·h-1 010 20 30 40 50 60 70 80 0 100 200 300 400 500 600 mean Hz Acco ding o GAVRIELI and co-wo ke s [12, 13], col- lapsible ubes may show low limi a ion and hen high equency oscilla ions o wo di e en ypes: he walls o he ube may oscilla e wi hou comple e collapse; o he e may be oscilla ion wi h a comple e closu e in each cycle o oscilla ion. Oscilla ion wi hou comple e clo- su e co esponds o he sound o a wheeze, whe eas oscilla ion wi h comple e closu e would co espond o he sound o sno ing. In he p esen s udy, wo di e - en ypes o sound spec a we e ound. The i s ype, cha ac e ized by a undamen al equency and ha mon- ics, was simila o he spec um obse ed du ing speech, especially emission o owels [14]. I was cha ac e ized by a high in ensi y peak co esponding o he unda- men al equency, ollowed by a se ies o highe e- quency peaks wi h he spec um occupying a b oad band o equency ha monics, ( ig. 1). The second pa e n o sound spec um was cha ac e - ized by a low equency peak, wi h he es o he ene gy o he sound spec um sca e ed a ound a na owe ange o equencies, wi hou clea ly iden i iable ha monics ( ig. 2). This second pa e n could co espond o wha has been called " elaxa ion oscilla ions" o "milking" oscilla ion ubes [15]. In he p esen s udy, di e ences be ween peak e- quency o bo h g oups ( able 3) we e obse ed. Al hough his esul was signi ican , in e pa ien a iabili y was high due o cha ac e is ics o peak equency. Mean e- quency was highe in sno ing pa ien s bu no s a is i- cally signi ican . The p esen da a show subs an ial di e ences wi h espec o he esul s o PEREZ-PADILLA e al. [4]. In hei s udy, he peak equency in pa ien s wi h simple sno - ing was lowe han in he p esen s udy, whe eas he peak in pa ien s wi h OSA was highe . Mo eo e , in pa ien s wi h OSA, hey also ound a esidual powe a equencies a ound 1,000 Hz. Ne e heless, he esul s o bo h s udies show some simila i ies: simple sno - ing was cha ac e ized in bo h s udies by a undamen- al equency and he p esence o ha monics ha a ely exceeded 500 Hz. In pa ien s wi h obs uc i e sleep apnoea, ha monics a e much mo e di icul o iden i y. In he p esen s udy, he second b oade peak equen- cy abou 1,000 Hz was no iden i ied. The main di e ence be ween he p esen s udy and he s udy by PEREZ-PADILLA e al. [4], a leas as con- ce ns sno ing, es s on he me hod and eco ding equip- men . They placed hei mic ophone on he manub ium s e ni, whe eas we placed ou mic ophone abo e he la - ynx. The il e ing e ec o he uppe ai way ca i ies is qui e di e en om ha o he achea and main b onchi, necessa y o he sound o each a mic ophone placed on he manub ium s e ni. I is, hus, possible ha he di e ences in he eco ding equipmen may be espon- sible o he quan i a i e di e ences be ween he p e- sen esul s and hose o PEREZ-PADILLA e al. [4]. The di e ence in se e i y o he OSA be ween ou pa ien s and hei s p obably canno explain he di e ences be ween indings. Indeed, in he p esen s udy a sig- ni ican co ela ion was ound be ween he se e i y o he OSA synd ome exp essed as he AHI and a ious indices desc ibing he noise spec um. This co ela ion implies ha , he mo e se e e he OSA, he lowe he equencies desc ibing he spec um. The e o e, one would expec he pa ien s o PEREZ-PADILLA e al. [4]. who had a mo e se e e OSA synd ome, o ha e e en lowe equencies han he pa ien s in he p esen sudy. I is known ha in OSA he e is oedema o he so pala e [16]. I can be hypo hesized ha , he highe he AHI he mo e se e e he oedema, wi h a co esponding inc ease in he mass o he so pala e. Vib a ing s uc- u es emi a sound spec um which is ela ed o hei mass, in such a way ha , he highe he mass he lowe he equency spec um o he sound [10]. This could explain he p esen esul s bo h in e ms o he lowe equencies in OSA pa ien s compa ed wi h pa ien s wi h simple sno ing, and in he nega i e co ela ion be ween he se e i y o he AHI and he sound spec um. Though he gene al spec a pa e ns a e di e en in OSA pa ien s and simple sno e s, he wo pa ien s wi h he leas se e e o m o OSA showed sound pa e ns including ha monics. By con as , only one pa ien wi h OSA showed a peak equency highe han 150 Hz, and only one simple sno e showed equencies o less han 150 Hz. An impo an ca ea should be bo n in mind con- ce ning he p esen s udy. In pa ien s wi h OSA, only he i s h ee b ea hs a e a comple e apnoea we e analysed. Those h ee b ea hs gene ally ake place du - ing he a ousal ha ends each apnoea. Sleep apnoea is p ecisely cha ac e ized by he occu ence o apnoeas ollowed by an a ousal ha allows esump ion o low. Thus, we eel i jus i ied o ha e pe o med he p esen s udy in his way. I u u e s udies con i m hese es- ul s, hen au oma ed acous ic analysis o sno ing could become a use ul ool in clinical sleep medicine o sepa- a e indi iduals wi h apnoea om hose wi h simple sno - ing. In conclusion, we ha e ound signi ican di e ences in he sound spec um o sno es be ween pa ien s wi h simple sno ing and pa ien s wi h obs uc i e sleep apnoea. These signi ican di e ences we e obse ed despi e he p esence o some o e lap be ween he wo g oups. Mo e- o e , he e was a signi ican co ela ion be ween indices desc ibing he sound spec um and he se e i y o he obs uc i e sleep apnoea synd ome. Re e ences 1. Luga esi E, Ci igno a F, Coccagna C, Piana C. Some epidemiological da a on sno ing and ca dio espi a o y dis u bances. Sleep 1980; 3: 221–224. 2. No on PG, Dunn EV. Sno ing as a isk ac o o dis- ease: an epidemiolocial su ey. B Med J 1985; 291: 630–632. 3. Dalmasso F, Benede o G, Pogolo i R, Righini G, Spagnolo R. Digi al p ocessing o sno ing sounds. Eu Respi J 1990; 3: 528–532. 4. Pe ez-Padilla JR, Slawinski E, Di ancesco LM, Feige RR, Remme s JE, Whi elaw WA. Cha ac e is ics o he sno ing noise in pa ien s wi h and wi hou occlu- si e sleep apnea. Am Re Respi Dis 1993; 147: 635–644. 5. Ame ican Tho acic Socie y. S anda diza ion o spi o- me y; 1987 upda e. Am Re Respi Dis 1987; 136: 1285–1298. 6. Roca J, Sanchis J, Agus i-Vidal A, e al. Spi ome ic e e ence om a Medi e anean popula ion. Bull Eu Physiopa hol Respi 1986; 22: 217–224. ACOUSTIC ANALYSIS OF SNORING 2369 7. Rech s a en A, Kales A (eds). A manual o s anda d- ized e minology, echniques and sco ing sys em o sleep s ages o human subjec s. Be hesda, Na ional Ins i u es o Heal h, 1968. 8. G o be g JB, Ga iely N. Flu e in collapsible ubes: a heo e ical model o wheezes. J Appl Physiol 1989; 66: 2262–2273. 9. Liis o G, S anescu DC, Ve i e CL, Rodens ein D, Aube -Tulkens G. Cha ac e is ics o sno ing in obs uc- i e sleep apnoea pa ien s and heal hy sno e s du ing sleep (Abs ac ). Eu Respi J 1989; 2: 330. 10. Liis o, S anescu DC, Ve ie C. Pa e n o simula ed sno ing is di e en h ough mou h and nose. J Appl Physiol 1991; 70: 2736–2741. 11. Liis o G, S anescu DC, Ve ie C, Rodens ein DA, Aube -Tulkens G. Pa e n o sno ing in obs uc i e sleep apnea pa ien s and in hea y sno e s. Sleep 1991; 14: 517–525. 12. Ga iely N, Pal i Y, Al oy G, G o be g JB. Measu emen and heo y o wheezing b ea h sounds. J Appl Physiol: Respi a En i on Exe cise Physiol 1984; 57 (2): 481–482. 13. Ga iely N, Jensen O. Theo y and measu emen s o sno es. J Appl Physiol 1993; 74(6): 2828–2837. 14. Fiz JA, Mo e a J, Abad J, e al. Acous ic analysis o owel emission in obs uc i e sleep apnea. Ches 1993; 104: 1093–1096. 15. Cancelli C, Pedley TJ. A sepe a ed- low model o col- lapsible- ube ocilla ions. J Fluid Mech 1985; 157: 375–404. 16. Ryan CF, Lowe AA, Li D, Flee ham JA. Magne ic es- onance imaging o he uppe ai way in obs uc i e sleep apnoea be o e and a e ch onic nasal p essu e he apy. Am Re Respi Dis 1991; 144: 939–944. J.A. FIZ ET AL. 2370