THRESHOLD PRESSURE FOR VOCAL FOLD COLLISION

Transcription

1 THRESHOLD PRESSURE FOR VOCAL FOLD COLLISION (Tröskeltryck för stämbandskollision) by Laura Enflo Subject: Musical Acoustics (musikakustik) Supervisor: Prof. Em. Johan Sundberg Examiner: Prof. Sten Ternström Stockholm, April 12, 26 Department of Speech, Music and Hearing School of Computer Science and Communication KTH, Royal Institute of Technology, Stockholm, SWEDEN

2 Acknowledgments This Master s thesis has been carried out at the Helsinki University of Technology (TKK), Laboratory of Acoustics and Audio Signal Processing, Espoo, Finland, in cooperation with my home university, the Royal Institute of Technology (KTH), Department of Speech, Music and Hearing, Stockholm, Sweden, between September 25 and March 26. The work was partially funded by grants from the EU (Socrates/Erasmus) and TKK. I would like to express my gratitude and thanks to my supervisor Professor Emeritus Johan Sundberg, KTH, for his help and encouragement and to Professor Paavo Alku, TKK, for his support and for giving me the opportunity to write my Master s thesis for KTH at TKK. I would like to thank Ph.D. students M.A. Laura Lehto, M.Mus Eva Björkner and M.Sc. Heidi-Maria Lehtonen for their assistance, companionship and patience while sitting in the same room, M.A. Patty Huang for reading through the English part of my thesis and suggesting corrections and M.A. Laura Lehto and Ph.D. Tom Bäckström for helping me with the abstract in Finnish. It has been a pleasure to be a part of the friendly working atmosphere in the Acoustics Laboratory, many thanks to all of you. In addition, I would like to thank my teachers in singing and other areas of the music field and my lecturers at KTH, who have generously given me a part of their knowledge. You make such important and good work, too little acknowledged. I especially think of M.Mus Britta Sundberg, M.Mus Gun Blüchert, M.Mus James Leatherbarrow, Ph.D. Svante Granqvist and Asso. Prof. Kirsti Mattila. I want to thank all of the singers who volunteered as subjects for the measurements at KTH in Stockholm. You kindly played or rather sang an important role in the making of this work. Last but not least, I want to acknowledge all of my friends and family for their care and inspiration, together with my late grandmother Brita and my late grandfathers Anton and Sakari. Each and every one of you makes and made a difference. Otaniemi, , Laura Enflo

3 Table of Contents Abstract Sammanfattning Tiivistelmä List of Abbreviations 1. Introduction 1 Subglottal Pressure 1 Phonation Threshold Pressure 1 Vocal Warm-Up 2 Collision Threshold Pressure 3 2. Experiment 3 3. Instruments 4 4. Analysis 5 5. Results and Discussion 7 CTP subtracted from PTP 7 CTP and Vocal Warm-Up 9 PTP and Vocal Warm-Up 1 Reproducibility Problems Conclusions and Further Work 15 References 17 Appendix: Table 3 18

4 Abstract A new measurement has been investigated, called the collision threshold pressure (CTP), meaning the minimal pressure required to initiate vocal fold collision. The purpose was to find a measurement that would be more accurate and less time-consuming to collect than the presently used phonation threshold pressure (PTP). Fourteen amateur singers six female and eight male were measured using an electroglottograph (EGG) and a pressure transducer that the subjects held in the corner of their mouths. The EGG showed if the vocal folds collide. All subjects were measured at several pitches before and after vocal warm-up. Both CTP and PTP tended to be higher before vocal warm-up. In addition, the differences varied greatly both within and between subjects. A clear result was that the CTP analysis demanded less time than the PTP analysis. The standard deviation of the PTP values collected here were smaller than those published in a previous investigation. Moreover, the standard deviations for CTP and PTP were compared in two subjects and were found to be lower for CTP. Thus, CTP may be a more useful measurement of vocal fold characteristics than PTP. Keywords: collision threshold pressure (CTP), phonation threshold pressure (PTP), vocal warm-up, electroglottography (EGG), singing, subglottal threshold pressure Sammanfattning Ett nytt mått har undersökts, kallat kollisionströskeltrycket (CTP), dvs det lägsta tryck som krävs för att stämbanden ska kollidera med varandra. Syftet var att hitta ett mått som är mer exakt och som tar mindre tid att mäta än det nuvarande måttet, fonationströskeltryck (PTP). Fjorton amatörsångare, sex kvinnor och åtta män, deltog i experimentet, där en elektroglottograf (EGG) och en tryckmätare, som placerades i ena mungipan, användes. EGG:t visade om stämbanden kolliderar med varandra. Alla försökspersoner mättes vid flera olika tonhöjder före och efter uppsjungning. Både CTP och PTP tenderade att vara högre innan rösten blivit uppvärmd. De individuella skillnaderna var dock stora, även mellan olika tonhöjder för varje försöksperson. Ett tydligt resultat var annars att analysen av CTP krävde mycket mindre tid än analysen av PTP. Standardavvikelsen för PTP-värdena var lägre i det här experimentet jämfört med värden från en äldre undersökning. Dessutom jämfördes standardavvikelserna för CTP och PTP med varandra för två försökspersoner och standardavvikelsen för CTP var lägre. CTP kan alltså vara ett mer användbart mått än PTP för stämbandskarakteristik. Sökord: kollisionströskeltryck (CTP), fonationströskeltryck (PTP), uppsjungning, elektroglottografi (EGG), sång, subglottalt tröskeltryck

5 Tiivistelmä Tässä työssä esitellään uusi tapa mitata subglottaalista kynnyspainetta. Tätä kutsumme yhteentörmäyskynnyspaineeksi (CTP). Tarkoitus oli löytää mittari, joka on täsmällisempi ja ajankäytöllisesti tehokkaampi kuin nykyinen mittausmenetelmä, fonaatiokynnyspaine (PTP). 14 koehenkilöä, kahdeksan miestä ja kuusi naista, tutkittiin käyttämällä elektroglottografiaa (EGG) ja suupieleen sijoitettua painemittaria. EGG näytti milloin äänihuulet saavuttavat sulkuvaiheen. Mittaukset tehtiin ennen äänen lämmittelyä ja sen jälkeen, ja osoittautui, että CTP ja PTP olivat keskimäärin suurempia silloin kun ääntä ei oltu lämmitelty. Yksillöiset erot vaihtelivat suuresti, myös saman henkilön eri äänenkorkeuksia mitatessa. Ilmeistä oli kuitenkin, että CTP oli suurempi kuin PTP kaikilla koehenkilöillä ja sävelkorkeuksilla, ja CTP:n analyysi vei vähemmän aikaa kuin PTP:n analyysi. CTP:n keskihajonta oli pienempi kuin PTP:n keskihajonta niillä kahdella koehenkilöllä, joilla oli riittävä joukko mittausarvoja. Tämän tutkimuksen PTP-arvoja verrattiin erään aiemman tutkimuksen PTP-arvoihin. Uuden tutkimuksen PTP-arvot olivat luotettavampia kuin vanhan tutkimuksen. CTP-arvot osoittautuivat kuitenkin kaikista luotettavimmiksi. Hakusanoja: yhteentörmäyskynnyspaine (CTP), fonaatiokynnyspaine (PTP), äänen lämmittely, elektroglottografia (EGG), laulu, subglottaalinen kynnyspaine List of Abbreviations CTP DSP EGG MV PTP SD VFCA collision threshold pressure digital signal processing electroglottogram/electroglottography mean value phonation threshold pressure standard deviation vocal fold contact area

6 1. Introduction Subglottal Pressure In singing, breathing habits are considered essential. The vocal folds are brought to vibration by the air pressure in the lungs. Such vibration is necessary to produce phonation. The needed airflow originates from the lungs and passes through the glottis, the V-shaped opening between the vocal folds. The subglottal pressure controls vocal loudness and must be adjusted to fundamental frequency and glottal adduction. There are also other factors that are not yet completely understood. The subglottal pressure is defined as the pressure by which the air pressure in the lungs exceeds atmospheric pressure 1. The most direct method of measuring of subglottal pressure is to insert a needle into the trachea through the tissues below the cricoid cartilage 1. A few singers have been measured in this way. An alternative, indirect method of measurement is now widely used 2. The subglottal pressure can be estimated by measuring the oral pressure during /p/ occlusions, since /p/ is an unvoiced, occlusive consonant. This means that /p/ is produced by abducting the vocal folds and simultaneously blocking the lip opening and then opening it suddenly to let out a single burst of air 3. During the /p/ occlusion the pressure in the mouth cavity is equal to the subglottal pressure, since the glottis is open and the lips are closed. To be able to measure the subglottal pressure for a specific pitch in this way, the subjects need to repeatedly pronounce a vowel following the consonant /p/. Often the vowel /a/ or / æ/ is used. An advantage of using these vowels is that the tongue position is low and that the first formant (73-13 Hz depending on gender and age) 1 is mostly higher than the fundamental frequency. As an example, a high first formant facilitates inverse filtering for the purpose of voice source analysis in terms of flow glottograms. In this work, the term subglottal pressure will be used for the pressure measured during the oral occlusion for the consonant /p/. Phonation Threshold Pressure Ingo Titze introduced the concept of phonation threshold pressure (PTP), i.e. the minimal lung pressure required to initiate and sustain vocal fold oscillation, which has been found to reflect significant characteristics of the vowels. In 1988, Titze derived an equation based on theory that showed how PTP increases with pitch as the vocal-fold thickness decreases: Equation 1: PTP = k t Bcξ / T where k t is a transglottal pressure coefficient, B is the mean damping coefficient for mechanical vibration in the tissue, c is the mucosal wave 1

7 velocity in the vocal fold cover, ξ is the prephonatory glottal half-width and T is the thickness of the vocal folds 4. Titze applied Equation 1 to results from models, and found that the best fit was obtained in using an equation with the following constants: Equation 2: PTP = (F / F M ) 2 where PTP is measured in cm H 2 O, F is the fundamental frequency and F M is the mean value for conversational speech (19 Hz for females and 12 Hz for males) 4. The same mean values have been used for singing as well 5. A low PTP is assumed to correspond to physiologically more efficient phonation and reduced phonatory effort, and hence PTP measurements have been assumed to have many potential areas of use. For example, it has been used successfully as a tool in documenting vocal fatigue (Chang & Karnell 6, Solomon & Stemmle DiMattia 7 and Milbrath & Solomon 8 ), where the PTP was found to increase during vocal fatigue. The PTP has also been used in looking for the effect of laryngeal lubricants including different kinds of medication and water. Three studies, with four subjects each, reported contradictory results (Solomon & Stemmle DiMattia 7, Verdolini et al. 9 and Solomon et al. 1 ) and one study by Roy et al. 11, with eighteen female subjects, reported that a lowering of PTP was seen for twenty minutes after the use of Mannitol, an agent that stimulates osmotic water flux to the luminal airway surface. However, no other substances, including water, were found to affect the PTP in that study. Three investigations have analysed the effects of vocal warm-up on the PTP as shown in the Vocal Warm-Up section below. The measurement of PTP is complicated by the difficulty for human beings to produce the softest possible sound. As a consequence, the values of PTP measurements are often reduced by a wide scatter. In addition, the analysis is very time-consuming. For those reasons, it is desirable to find a better measure of subglottal pressure of relevance to vocal fold properties. Vocal Warm-Up Many singers prefer to warm up their voices before singing, both professionals (Fleming 12 and Nilsson 13 ) and amateurs. Vocal warm-up is considered helpful for singers in order to use their whole voice and to avoid injuries while singing. A pilot study made by Elliot, Sundberg and Gramming 5 in 1995 showed that there is a significant difference in the PTP level before and after the voice has been warmed up. However, the results differed from subject to subject and failed to provide a definitive answer to the question if a vocal warm-up really lowers the PTP. In 23, Motel et al. 14 found that sopranos tended to have a higher PTP for high frequencies after vocal warm-up, but at other frequencies no clear effects were seen. In the same year, Milbrath & Salomon 8 made a study to see whether vocal warm-up helped persons who reported chronic vocal fatigue to 2

8 get a better voice function during and after one hour of loud reading. However, the results showed high variability between subjects. The phonatory effects of vocal warm-up have also been investigated using other measurements than PTP. In 1994, Stemple et al. 15 found significant changes in phonation volume, flow rate, maximum phonation time, and frequency range among female subjects with normal voices after four weeks of warm-up exercises compared to subjects who had not had any warm-up at all. In 21, Vintturi et al. 16 found changes mainly in ergonomic factors, which, however, differed depending on gender and type of phonation. In 25, Amir et al. 17 observed acoustical effects of vocal warm-up in twenty young female singers. They found that vocal warm-up increased the amplitude of the singer's formant and that it improved noise-to-harmonic ratio. It also reduced frequency-perturbation and amplitude-perturbation values. Collision Threshold Pressure A problem with PTP is that it concerns very low pressures. If the PTP for a given pitch is, say, 2 cm H 2 O, a 1 cm H 2 O of measurement error corresponds to a 5% variation. For this reason it seemed worthwhile to explore the possibility of finding an alternative to PTP. For very low subglottal pressures, i.e. in very soft phonation, the vocal folds fail to reach contact. The folds vibrate, but with an amplitude so small that the folds never collide. If the subglottal pressure is increased, however, vibration amplitude is increased such that the folds start to collide. The minimal pressure required to initiate vocal fold collision, henceforth the collision threshold pressure (CTP), is a new measure which seemed likely to be related to vocal fold properties. Being higher than the PTP, it seemed possible that subjects would find it easier to intentionally produce and replicate CTP than PTP and, hence, that it could be determined with a higher accuracy. In addition, the CTP could possibly require less time to analyse. The purpose of the present investigation was to explore this new measure. 2. Experiment Fourteen subjects, six female and eight male, participated in the experiment. All of them were amateur singers with a considerable amount of vocal training. The reason why amateur singers were chosen as subjects was that the warm-up effect could be assumed to be greater for such subjects who usually do not sing as frequently as persons who earn their living from singing. Each subject was recorded twice: before and after vocal warm-up. The warm-up was carried out in two ways. Nine subjects warmed up their voices individually according to their own habitual procedures, without instructions from the experimenter. This warm-up took about 2-3 minutes. For the remaining five subjects, two female and three male, the warm-up consisted of 1-2 hours of choir rehearsal. Every subject felt warmed up before the second recording. 3

9 To obtain an estimate of reproducibility of the PTP and CTP measuring methods two of the subjects, one female and one male, were recorded on three different occasions. The experimental set-up is illustrated in Fig. 1, page 5. Electroglottography (EGG) was used to identify vocal fold collision. An electroglottograph is provided with two electrode plates, which are placed on each side of the larynx. The idea with EGG is to send an electrical signal from one electrode to the other and record the amplitude of this signal. When the vocal folds are closed, the signal can pass, such that the amplitude is high, and when the glottis is open, the amplitude is low. Contact gel of the same type as that used during ultrasound examinations was spread on the surfaces of the electrodes in order to improve the skin contact. The recordings were made in a studio with the subject sitting on a chair with the electroglottograph tied around his/her neck, holding a catheter provided with a pressure transducer in the corner of his/her mouth. The distance from the subject s mouth and the microphone was 3 cm. Each pitch was chosen according to the normal voice range of each subject, and was given to the subjects from a singing synthesizer program (Madde). The subjects were instructed to repeat the syllable [pa:] in a decrescendo on each pitch, starting at medium loudness and continuing until voicing ceased. Each sequence was repeated three to six times, in order to obtain at least three measurable sequences. Subglottal pressure was calibrated by putting the catheter provided with a pressure transducer at the bottom of a bottle filled with water. The subglottal pressure was measured in centimeters of water, where 1 cm H 2 O =.1 kpa =.1 atm. 3. Instruments A condenser microphone (Brüel & Kjær type 43), was used for the audio recordings. The microphone pre-preamp and power supply (B&K type 2812) was set to a gain of db. In order to amplify the audio signal, it went through a DSP Audio Interface Box (Nyvalla DSP, microphone preamp), before entering the computer. The EGG was measured with a two-channel electroglottograph (Glottal Enterprises, model EG 2), with settings VFCA (Vocal Fold Contact Area) and a low frequency limit of 4 Hz. Before entering the computer, the signal was examined on a digital storage oscilloscope (Gould Advance model OS4). The subglottal pressure was measured with a pressure transducer (Gaeltec Ltd. type 7b). 4

10 Loudspeaker Computer 1 2 CONTROL ROOM Oscilloscope STUDIO DSP Interface Box Electroglottograph Pressure Transducer Microphone Power Supply [pa: pa: pa: pa:] Fig. 1. Wiring diagram from the experiment. 4. Analysis The analysis was performed using the Soundswell Signal Workstation TM (Soundswell Core TM 4.) from Hitech Development AB, Sweden. As previously mentioned, the recordings were made on three channels: audio, EGG and subglottal pressure. The subglottal pressure channel (channel 2 in Fig.1) was processed for analysis according to the following procedure: 1. using the pressure calibration made during the recordings, the DC offset was identified and corrected such that zero volt corresponded to zero pressure 2. as the signal was inverted, it was multiplied by -1 5

11 3. calibration, so that the correct value (in cm H 2 O) appeared on the y- scale 4. low-pass filtering the signal with cut-off frequency F = 5 Hz, thus eliminating noise The EGG channel (channel 1 in Fig. 1) required processing as well. This comprised: 1. high-pass filtering with cut-off frequency F1 = 9 Hz, thus eliminating noise 2. full-wave rectifications of the signal. This facilitated comparing EGG amplitudes for determining vocal fold collision CTP was determined for each tone as sung by each subject. Three pressure values on each side of the threshold were measured. These values were identified with the aid of the EGG signal. The subglottal pressures appearing closest to the disappearance or a steep decrease of the EGG amplitude were accepted as lying just below the CTP. Likewise, the subglottal pressures appearing closest to the reappearance or a steep increase of the EGG amplitude were accepted as lying just above the CTP. Then, the average was calculated from each pair of pressures appearing around the collision threshold. In this way, three estimates of the CTP were obtained. Finally, the average of these estimates was calculated. All computations were done with Microsoft Excel. Standard deviations of CTP values were calculated for the pitches sung by all subjects. Paired t-tests were made in order to find out the probability that the subglottal pressure values would not change after vocal warm-up for each subject and pitch. PTP analysis was difficult, and only one PTP value could be collected per pitch for every subject. The reason was that in many cases the subjects stopped repeating the syllable [pa:] before voicing ceased. Standard deviations of PTP values were calculated for the pitches sung by the two subjects who had been recorded in more than one session, that is, LF1 and JM2. Two-sample, heteroscedastic t-tests were made in order to find out the probability that the two compared recordings for each of the subjects JM2 and LF1 would have the same subglottal pressure values for each pitch. The results are shown in Table 1, page 13. The PTP values collected from all subjects were compared with those computed by Equation 2. For each pitch and subject, the difference between the observed PTP value and the PTP value predicted by Equation 2 was determined. Then, the average of these differences was calculated across all pitches sung by each subject. These averages are listed together with their standard deviations in Table 2, page 15 and Table 3, Appendix. These means 6

12 and standard deviations can be compared with those reported in a previous investigation of the effects of vocal warm-up on PTP 5. The pitches were measured and expressed in the logarithmic semitone unit, thus facilitating comparison between male and female subjects. The tone C2 was used as reference. In the following graphs, subglottal pressure was also plotted using a logarithmic scale. For the subjects a code was made where the first letter would stand for the beginning letter of the first name and the second letter would refer to the subject s gender female (F) or male (M). In those cases where the first letter was the same for several subjects, a number was added so that they could be separated from each other. 5. Results and Discussion Subject EF3 had to be eliminated from analysis since 1) no contact gel was used which reduced the quality of the EGG-signal and 2) the catheter provided with a pressure transducer accidentally moved into a position between the lips. The latter problem occurred for many other subjects, but luckily at just a few places in the sound files, except for EF3 and LF1 recording 3, who both had to be eliminated from analysis. No particular subglottal pressure value or EGG amplitude differences between the self-warmed subjects and the ones who visited a choir rehearsal could be found. Hence all subjects were pooled in one group. Two of the subjects (FM and JM2, recording session no 2) reported having a cold. For those two, the CTP values were lower than for most of the other subjects. That is too few subjects to allow any potential conclusions. However, during a cold the vocal folds become swollen and have more mass than normally, which explains why singers with a cold sometimes are flat in their intonation 2. Consequently, it should demand less effort to sing low notes. It has also been postulated that PTP should increase when the vocal fold thickness decreases (see Equation 1). Hence, one might assume that the CTP decreases below normal values for low pitches, when the vocal folds are thicker than usual, because of an illness or for other reasons. CTP subtracted from PTP Obviously CTP shows higher values than the PTP, regardless of pitch and subjects. Thus, CTP-PTP > for both females and males (Fig. 2 & 3). 7

13 CTP Minus PTP, All Females C T P -P T P [c m H 2 O ] Semitones (F re C2) Warmed Not Warmed Fig. 2. The difference between the CTP and the PTP for all female subjects individually and for each pitch. CTP Minus PTP, All Males C T P - P T P [c m H 2 O ] Semitones (F re C2) Warmed Not Warmed Fig. 3. The difference between the CTP and the PTP for all male subjects individually and for each pitch. The difference between the CTP and the PTP is considerable between individuals for each pitch. 8

14 CTP and Vocal Warm-Up When it comes to the CTP, a difference can be seen before and after vocal warm-up. CTP Warmed and Not Warmed, All Females Subg lo ttal Pressure [log cm H2O] 1,4 1,2 1,8,6,4, CTP Warmed CTP Not Warmed Semiton es (F re C2) Fig. 4. The CTP for all females individually and for each pitch, before and after vocal warm-up. CTP Warmed and Not Warmed, All Males 1,2 Subglottal Pressure [log cm H2O] 1,8,6,4, CTP Warmed CTP Not Warmed Semitones (F re C2) Fig. 5. The CTP for all males individually and for each pitch, before and after vocal warm-up. The difference in CTP before and after vocal warm-up is not the same for all subjects. Paired t-tests showed that the subglottal pressure values would not 9

15 change after vocal warm-up for each pitch with a probability of.1 for all males and.28 for all females. The ratios for CTP before and after vocal warm-up were computed to show the individual variations, see Fig. 6. Ratio CTP After/Before Warmup For All Subjects R a t i o C T P W a r m e d / C T P N o t W a r m e d 2 1,8 1,6 1,4 1,2 1,8,6,4, Semitones (F re C2) BM EM EF1 EF2 FM JM1 JF JM2 LM LF1 LF2 OM PM Fig. 6. Ratio between CTP after and before warm-up, for each pitch and subject. All subjects included, both female and male. In general, the CTP is lower after vocal warm-up according to Fig. 4 & 5, but the individual differences are great, also between pitches per individual. A little more than half of the ratio points are below the 1-line in Fig. 6, which means that the CTP was generally higher before vocal warm-up. It seems that vocal warm-up has a clear effect on all singers since most of the points are not close to y = 1. Only, this effect varies a lot between pitches and subjects. PTP and Vocal Warm-Up In the PTP a difference in subglottal pressure can also be seen before and after vocal warm-up, according to Fig. 7 & 8. The measured data fit reasonably the calculated by means of Titze s Equation 2, see also Table 2, page 15 and Table 3, Appendix. 1

16 PTP Warmed and Not Warmed, All Females S u b g lo t ta l P r e s s u r e [l o g c m H 2O ] 1,4 1,2 1,8,6,4,2 -,2 -,4 -, Semitones (F re C2) PTP Warmed PTP Not Warmed Titze's Empirical PTP Fig. 7. The PTP for all females individually and for each pitch, before and after vocal warm-up. PTP Warmed and Not Warmed, All Males S u b g lo tta l P r e s s u r e [lo g c m H 2 O ] 1,2 1,8,6,4,2 -,2 -,4 -, Semitones (F re C2) PTP Warmed PTP Not Warmed Titze's Empirical PTP Fig. 8. The PTP for all males individually and for each pitch, before and after vocal warm-up. To show the individual differences before and after vocal warm-up, a ratio was computed for PTP (Fig. 9) in the same way as for CTP (Fig. 6). No inclinations for the ratio to differ much from the CTP ratio could be seen. Paired t-tests showed that the subglottal pressure values would not change after vocal warm-up for each pitch with a probability of.45 for all males and.3 for all females. 11

17 Ratio PTP After/Before Warmup For All Subjects R a tio P T P W a rm e d / P T P N o t W a rm e d 4 3,5 3 2,5 2 1,5 1, Semitones (F re C2) BM EM EF1 EF2 FM JM1 JF JM2 LM LF1 LF2 OM PM Fig. 9. The PTP after vocal warm-up divided by the PTP before the voice has been warmed up, for each pitch and subject. All subjects included, both female and male. Reproducibility How good is the reproducibility of CTP and PTP at each pitch? The standard deviations for the CTP were computed for all subjects and the PTP was calculated for JM2 and LF1, who each had comparable data from two of the three recording sessions, since JM2 had a cold during the second recording session and the third recording session of LF1 had to be eliminated due to noise. All standard deviation values in the plots are mean values of the standard deviations for each pitch and each subject. The standard deviation values for the CTP were generally lower than for the PTP, see Fig. 9 & 1. SD, CTP Versus PTP, LF1, Recording 1 & 2 1,4 Standard Deviation [cm H2O] 1,2 1,8,6,4, CTP PTP Semito nes (F re C2) Fig. 9. Female subject LF1: Standard deviation for CTP and PTP, respectively. 12

18 SD, CTP Versus PTP, JM2, Recording 1 & 3 Stand ard Deviation [cm H2O] CTP PTP Semitones (F re C2) Fig. 1. Male subject JM2: Standard deviation for CTP and PTP, respectively. Two-sample, heteroscedastic t-tests showed the probability that the subglottal pressure values for the two compared recordings for each of the subjects JM2 and LF1 would be the same for each pitch. The results are shown in the table below (Table 1). Subjects CTP Not Warmed CTP Warmed PTP Not Warmed PTP Warmed JM LF Table 1. The probability that the subglottal pressure values from the two compared recordings (i.e. two sets of subglottal pressure values) for each subject and pitch would be equal. In view of the fact that there were not enough PTP values available for calculating the standard deviations for all subjects, only the standard deviations for the CTP were checked for all females and males, respectively. It can be seen from Fig. 11 & 12 that the standard deviations for LF1 and JM2 do not seem to differ much from the standard deviations for the other subjects. 13

19 Standard Deviation [cm H2O] 1,9,8,7,6,5,4,3,2,1 SD, CTP, All Females Versus LF Semitones (F re C2) MV All Females except LF1 MV LF1 Fig. 11. Comparison between standard deviations for LF1 and standard deviations for the other female subjects. SD, CTP, All Males Versus JM2 Standard Deviation [cm H2O] 1,9,8,7,6,5,4,3,2, MV All Males except JM2 MV JM2 Semitones (F re C2) Fig. 12. Comparison between standard deviations for JM2 and standard deviations for the other male subjects. It seems that CTP can be determined with lower standard deviation values than PTP. Even when taken into account the fact that the CTP values come from a greater number of measured points, this could not completely explain why the standard deviation is lower for CTP. In a previously mentioned study 5 made in 1995 where only PTP was measured, the differences between the PTP values and Titze s empirical values (Equation 2) were calculated for all pitches pooled and a table with the average values, including the standard deviations, were included in the report. The values for all of the subjects in the present study were calculated in the same way and compared with the older data, to check if they were similar, and if the new PTP values for this reason could be worthy of a comparison with the CTP values. See Table 2 below. 14

20 Subjects pooled Measurement in this work Before vocal warm-up (cm H2O) After vocal warm-up (cm H2O) Females Mean SD Males Mean.9 1. SD.8.7 Measurement from 1995 Females Mean SD Males Mean SD Table 2. The mean difference and the standard deviation of the mean difference between the measured PTP values and Titze s theoretical PTP values before and after vocal warm-up for females and males. New data compared with previously published values. All pitches included. For more detailed information, see Table 3, Appendix. The values for the subjects in the present study were by no means exceptional. In addition, the mean values for standard deviation for the groups were better than for the recording made in Hence, the PTP values coming from this experiment were not bad according to what the standard deviations showed. This indicates that the standard deviation for PTP is higher than the standard deviation for CTP, which means that CTP might be a more reliable measurement. 6. Problems One issue was that the catheter provided with a pressure transducer was hard to keep in the right place in the mouth. This sometimes led to unrealistic values of the pressure. That was the main reason for discarding data in a few cases, see the Results and Discussion section. Another problem was that the EGG sometimes failed to produce a signal at high frequencies, where the larynx position is sometimes higher. The reason for this could be because the larynx was higher up than the electrodes, or because of falsetto singing, especially since none of the subjects were countertenors (who sometimes have a different way of singing in the falsetto register). Falsetto singing sometimes occurs also in female voices 2. In this register the vocal folds typically fail to collide, so no CTP can be found. 7. Conclusions and Further Work Measurements indicate that the standard deviation for the CTP is lower than for the PTP. Another advantage of the CTP method for measuring subglottal threshold pressure is that it is less time-consuming to measure than the PTP method. However, the PTP method is more reliable at high pitches since the larynx height and the falsetto singing are problematic when measuring the CTP. 15

21 The effect of vocal warm-up is considerable for all subjects, but with a high between-subject variability. Clear instructions to the subjects are important in order to prevent a lack of PTP values below the threshold and to avoid unrealistic pressure values caused by misplacements of the catheter provided with a transducer. Further research is necessary to find out whether these indications about CTP are correct. The sample of fourteen subjects is too small to allow conclusions. 16

22 References [1] Rossing, T. D., Wheeler, P. & Moore, R. (22) The Science of Sound, 3 rd edition. San Francisco, CA: Pearson Education, Inc. [2] Sundberg, Johan (1986) Röstlära, 2 nd edition. Stockholm: Proprius förlag. [3] Hall, Donald E. (1991) Musical Acoustics. Belmont, CA: Wadsworth, Inc. [4] Titze, Ingo R. (1991) Phonation threshold pressure: A missing link in glottal aerodynamics, J. Acoust. Soc. Am. Vol. 91, pp [5] Elliot, N., Sundberg, J. & Gramming, P. (1995) What Happens During Vocal Warm-Up?, Journal of Voice Vol. 9, No. 1, pp [6] Chang, A. & Karnell, M. P. (24) Perceived Phonatory Effort and Phonation Threshold Pressure Across a Prolonged Voice Loading Task: A Study of Vocal Fatigue, Journal of Voice Vol. 18, No. 4, pp [7] Solomon, N. P. & Stemmle DiMattia, M. (2) Effects of a Vocally Fatiguing Task and Systemic Hydration on Phonation Threshold Pressure, Journal of Voice Vol. 14, No. 3, pp [8] Milbrath, R. L. & Solomon, N. P. (23) Do Vocal Warm-Up Exercises Alleviate Vocal Fatigue?, J. of Speech, Language, and Hearing Research Vol. 46, pp [9] Verdolini, K., et al. (22) Biological Mechanisms Underlying Voice Changes Due to Dehydration, J. of Speech, Language and Hearing Research Vol. 45, pp [1] Solomon, N. P., et al. (23) Effects of a Vocally Fatiguing Task and Systematic Hydration on Men s Voices, Journal of Voice Vol. 17, No. 1, pp [11] Roy, N., et al. (23) An Evaluation of the Effects of Three Laryngeal Lubricants on Phonation Threshold Pressure (PTP), Journal of Voice Vol. 17, No. 3, pp [12] Fleming, Renée (25) The inner voice. New York: Penguin Group Inc. [13] Nilsson, Birgit (1995) La Nilsson. Stockholm: T Fischer & Co. [14] Motel, T., Fisher, K. V. & Leydon, C. (23) Vocal Warm-Up Increases Phonation Threshold Pressure in Soprano Singers at High Pitch, Journal of Voice Vol. 17, No.2, pp [15] Stemple, J. C., Lee, L., D Amico, B. & Pickup, B. (1994) Efficacy of Vocal Function Exercises As a Method of Improving Voice Function, Journal of Voice Vol. 8, No. 3, pp [16] Vintturi, J., Alku, P., et al. (21) Objective Analysis of Vocal Warmup With Special Reference To Ergonomic Factors, Journal of Voice Vol. 15, No. 1, pp [17] Amir, O., Amir, N. & Michaeli, O. (25) Evaluating the Influence of Warmup on Singing Voice Quality Using Acoustic Measures, Journal of Voice Vol. 19, No. 2, pp

23 APPENDIX Table 3 Subject Before Mean (cm H 2 O) SD (cm H 2 O) After Mean (cm H 2 O) SD (cm H 2 O) Females EF EF JF LF1, rec LF1, rec LF Mean SD Males BM EM FM JM JM2, rec JM2, rec LM OM PM Mean.9 1. SD.8.7 Measurement from 1995 Females GM EL GB PG HH AMHB JI Mean SD Males ST SH JS Mean SD Table 3. The mean difference and the standard deviation of the mean difference between the measured PTP values and Titze s theoretical PTP values before and after vocal warm-up for females and males. New data and values from a previous investigation. All pitches and subjects included. 18