8. Summaries and comments on individual


Paper I. "Blowing pressures in bassoon, clarinet, oboe and saxophone"

Blowing pressure is a major input parameter in wind instruments. In previous investigations, some data had been published on this topic (Bouhuys, 1964, 1968; Navrátil M & Rejsek K, 1968; Pawlowski & Zoltowski, 1985, 1987; Bak & Doemler, 1987; Cossette, 1993). However, there was a shortage of systematic measurements describing details of the relationship between pitch, blowing pressure and dynamic level. Also, some previous studies reported results collected from huge numbers of players, including professionals and students, presumably obscuring systematic dependencies between the three parameters mentioned. The aim of many of these studies was mainly to provide typical and maximal pressure values for clinical applications. The basic hypothesis tested was that a systematic relation exists between blowing pressures and the acoustic properties of the instrument and of the tones produced.

We recorded blowing pressures in the mouth cavity of two professional players of each of four reed woodwinds (Bb clarinet, alto saxophone, oboe, bassoon). The players performed three different tasks: (1) a series of isolated tones at four dynamic levels, (2) the same series with a crescendo-diminuendo tones and (3) ascending-descending musical arpeggio played legato at different dynamic levels (pp, mp, mf, ff). The results showed that, within instruments, the players’ pressures exhibited similar dependencies of pitch and dynamic levels. Between instruments, clear differences were found with regard to the dependence on pitch. As can be seen in Figures 14-15, 20-21, 26-27 and 32-33, each instrument presented characteristic curves, differing only slightly between players.

Vibrato was not an object of study in this project, and the players were asked to avoid it, since it could be assumed to be associated with variations of the blowing pressure. Yet, some players occasionally produced a vibrato, as can be seen in Figure 6, p 44. In that particular case, the average blowing pressure was approximately 13 cmH2O, while the amplitude of the vibrato oscillations varied between 1.5 and 3.0 cmH2O peak-to-peak. This considerable modulation (about 10-25%) was still associated with a rather moderate vibrato effect. For a greater vibrato and at higher dynamical levels and pitches, even larger pressure modulations can be expected. This issue was investigated in Paper II.

The investigation focused on isolated, sustained tones, crescendo-diminuendo tones and arpeggi at fixed dynamical levels. We might assume that similar patterns would be found for other notes in the range studied and that blowing pressures for intermediate notes can be estimated by interpolation (see Paper I, Appendix, Table A). However, the fingering and the instrument’s responsiveness do not necessarily vary continuously between adjacent tones, and this may demand differing pressure values. Tone variations occurring in the playing of real music may require much more complex pressure changes, see Figure 1, Paper V.

The Paper suggested that the proprioceptive functions in the respiratory system, which are responsible for the perception of stimuli by the abdominal, thoracic and lung receptors, are highly relevant to wind instrument playing. This was the main topic of Paper V.

The roles of the authors in this Paper were as follows: LF: Experimental design, recording, analysis of data, authoring the draft for the manuscript; JS: Consultation, assistance in experiment and in editing the manuscript.

Paper II. "Aerodynamic input parameters and sounding properties in naturally blown reed woodwinds"

This study can be considered a complementation of Paper I, which focused blowing pressure. The present paper corroborated and expanded the data from Paper I and added information regarding the airflow resistance of the reed. Also measurements of airflow were carried out, allowing estimation of typical demands on air supply in the different instruments - oboe, alto saxophone, bassoon, and clarinet.

Commercially available clarinet and saxophone reeds are usually specified with respect to "hardness", using an arbitrary scale ranging from 1 (soft) to 5 (hard). The "hardness" of a reed is defined by means of bending tests, i.e. in a way similar to that of measuring stiffness of materials in engineering (Vandoren Products Catalog, 1983). However, the "hardness" of a reed may be associated with a number of other properties as demonstrated by Fuks (1995) for clarinet reeds. Reed "hardness" is highly relevant to bassoon and oboe playing, and is generally adjusted by the players when they manufacture the reeds. The acoustical and aerodynamical correlates of the "hardness" have not been experimentally demonstrated. According to the general opinion among players greater "hardness" is associated with the need for higher blowing pressures and embouchure forces as well as with a "darker" tone quality.

Input aerodynamic parameters in reed woodwinds were measured when professionals played different pitches at three dynamic levels. Two reeds were used, one rated as a hard and another as a soft reed. The contrasting reeds in our experiment showed clearly different characteristics in terms of airflow, blowing pressure and resistance.

The tasks consisted of playing long sustained tones with and without vibrato. Audio and blowing pressure signals were recorded. Lung volume variations were indirectly measured by a spirometric procedure, which also showed the average air consumption, i.e. the mean airflow. The tones were played in a laboratory room and also in a calibrated reverberant chamber allowing estimation of radiated sound power. Average values for flow resistance, aerodynamical power, and mechanical efficiency were computed. Airflow and input power varied considerably between instruments and between the two reed types, but generally increased with sound level. Airflow systematically increased with blowing pressure and dynamical level in all instruments. The harder reeds required higher airflow, blowing pressure and tended to produce tones of higher SPL.

According to previous investigations the pressure-flow characteristic should follow a bell-shaped curve under conditions of constant embouchure tightness, see Figure 9, and only the falling part of the curve can be used in playing. Yet, our pressure measurements indicated that an increase of blowing pressure was associated with an increase of dynamic level. Thus, an increase of pressure seemed to produce an increase of airflow. This apparent discrepancy suggests that the player must reduce the embouchure tightness during a crescendo. If the player keeps a constant embouchure while increasing blowing pressure, the tone gets softer until eventually the reed simply closes.

Vibrato can be generated in different ways, as mentioned above. The investigation focused on an intense, or clearly audible vibrato. For vibrato tones produced on oboe, bassoon and saxophone, wide pressure oscillations were observed, on average 10 cmH2O for the oboe and bassoon, and reaching values of 20 cmH2O in some cases. These great undulations cannot be produced merely by the laryngeal mechanism, as claimed in previous studies, but must require participation of expiratory forces.

Paper IIIa. "Prediction of pitch effects from measured CO2 content variations in wind instrument playing"

Paper IIIb. "Prediction and measurements of exhaled air effects in the pitch of wind instruments"

The studies reported in Papers IIIa and IIIb departed from a pilot study about respiratory conditions under different degrees of physical exercise and the effect of CO2 in the lungs on maximal breathhold time. A capnometer, a device for measuring carbon dioxide percentage in the air (Bhavani-Shankar et al., 1995), was used for measuring CO2 variations during performance on the oboe and the clarinet. This is a key gas in the respiratory control; the human body contains receptors that continuously monitor the amount of that gas, regulating the optimum level of gas changes and the rhythm of ventilation (Staub, 1991; Shea et al., 1996; Piiper & Scheid, 1982; Klocke, 1982; Flume et al., 1996; Fernando & Saunders, 1995; Banzet et al., 1996).

We observed wide variations in the CO2 contents of exhaled air, usually starting from 2.5-3% and reaching up to 8.5% in extreme cases, after more than 50 seconds of playing without taking a new breath. This value exceeded those published as extremes in medical textbooks (e.g., Schmidt & Thews, 1983) and even in studies of the effects of breathhold during diving (Lin, 1987).

The effect of these great changes in gas composition on the intonation in these instruments was analysed. As mentioned, fundamental frequency depends on the sound speed in the air column in these instruments. Nedeerven (1969) calculated the effective sound speed for a flute, taking into account the effect of temperature and also the composition of the gas and assuming a CO2 average of 2.5%.

In Paper IIIa, the variations in exhaled O2 were assumed to vary linearly with time between 21% and 15%, the generally accepted range (Schmidt & Thews, 1983).

The study further assumed warmed up instruments, so that the temperature effect would not be interfering with the effect of gas composition. Also, it was postulated that the player started after a deep breath, inducing low values of CO2 and high values of O2. Simultaneous measurements of CO2 and O2 were realized in Paper IIIb.

Predicted and observed decreases of fundamental frequency amounted to 27 cents and 16 cents, respectively. The discrepancy might be accounted for by factors not considered in this paper. Thus, air humidity was assumed at a fixed 100% value, i.e., saturated vapour. As a complement, we present a calculation of the impact of humidity changes.

The idealised situation with a warmed up instrument is assumed, but with humidity increasing from ambient values to vapour-saturated air during the first moments of playing. As mentioned in Paper IIIa, sound speed c is calculated by equation 1.1,

an alternative equation, resulting from Boyle's law, is


P0 is the ambient pressure

r 0 is the gas density

g , the specific-heat ratio, i.e. specific heat at constant temperature divided by specific heat at constant volume

R , a gas constant

T is the temperature in K



R0 is the universal gas constant , 8312 J/(kg.K)

M is the average molecular wheight of the dry composite gas.

This formula yields the same numerical results as those obtained by formulas 1.1 and 1.2 in Paper IIIb. In accounting for the effect of humidity, a thermodynamic formula valid for the narrow range of variations in temperature and in gas compositions (e.g. Pierce, 1981) is:


cwet sound speed in the humidified gas,

cdry sound speed in dry gas,

H fraction of H2O molecules in the air


From thermodynamical tables (ASME, 1967) it can be calculated that for saturated vapour at 40°C, the value for H is 0.07, approximately (Pierce, 1981). For a variation from 40% ambient humidity to 100% inside an instrument, i.e. reaching 60%, the isolated contribution of humidity should be approximately:


According to Equation 5 the increase in sound speed due to humidity should amount to 0.67%, or 12 cents in fundamental frequency for the sounds produced. This amount should oppose the expected maximum shift of -27 cents, due solely to the CO2 and O2 changes. The net shift of -15 cents seems realistic, as compared to approximately -16 cents observed in Paper IIIb and shown in Figure 3.

As pointed out in Paper IIIa, the impact of the effect of the gas mixture should also indirectly depend on the airflow. Paper II offered some experimental data regarding airflow through four reed instruments. The volume contained in an adult player’s upper airways, or the physiological dead space, amounts to approximately 150 ml. In cases of low airflow, it could take as much as 3 or 4 seconds for the intrapulmonary air to reach the reed. This would be the most critical phase for the gas effects on the intonation, as shown by Figures 2 and 3 in Paper IIIb. During that period, the quasi-ambient air would be fed to the instrument bore, and thereafter intrapulmonary air. In cases of high flow, the effect of the transient gas density will occur at the same time as other transitional processes, such as the onset of the tone, the adjustment of pitch and dynamic level, etc. Once the pulmonary air has filled the entire instrument a slowly changing effect on the fundamental frequency can be expected.

Papers IIIa and IIIb demonstrated that the discussed effect of gas changes may be a relevant factor in performance.

The outcomes of this study can be demonstrated by a simple experiment that may be useful in a music classroom. It requires a reed woodwind instrument, a manometer connected to the player’s mouth and an electronic tuner. After warming up, the subject will play a long steady tone at pp, keeping a fixed embouchure and blowing pressure as displayed by the manometer. Another observer will monitor the fundamental frequency as indicated by the tuner. The subject may blow a long tone interrupted by occasional brief pauses, without breathing. This is not expected to change the tuning. Then, the subject quickly exhales the remaining air in the lungs and resumes playing after a quick and deep breath. This time, an effect on tuning can be expected. This obviously is a simplified and less well-controlled version of the experiment described in Paper IIIb.

Paper IV. "Respiratory inductive plethysmography measurements on professional reed woodwind instrument players"

In Paper II, airflow in long sustained tones, played at different pitches and dynamic levels was measured. This, obviously, is a rather particular playing condition. To collect data from real performance a different method should be applied. As mentioned, the multiple factors in sound production in wind instruments make it a hard task to measure airflow directly, without severely affecting the system.

Respiratory Inductive Plethysmography (RIP) is a technique devised for clinical respiratory monitoring, introduced in the end of the 70's. It is still in frequent use as it is reliable, robust, non-invasive, intrinsically safe and represents a relatively low cost method (Chadha et al. 1982; Strömberg, 1996). The technique is based on the principle of self-inductance of a coil, which is connected to a high-frequency oscillator. Elastic bands are wrapped around the chest and abdomen of the subject; the bands are provided with an electric coil powered by the oscillating unit, see Figure 1 in Paper IV. The changes in the cross-sectional areas surrounded by the coils cause changes in the electromagnetic properties of the system. Thus, the individual variations of chest and abdominal volumes produce corresponding variations in the output signal of the RIP. The method is closely related to the two-degrees-of-freedom model for measurement of chest wall volume displacements (Konno and Mead, 1967; Loring and Bruce, 1986). According to this model, the variations in lung volumes are solely reflected in the variations of the two independent compartments, the rib cage and the abdomen, although a good accuracy can be attained only provided that there is no change in body position. The technique has been successfully applied to the analysis of respiratory movements in singers (Thomasson and Sundberg, 1997). In addition, studies have been carried out on the relation between respiration and phonation (Iwarsson et al., 1996). Regarding wind instrument playing, one single investigation (Cugell, 1986) has been published, particularly considering brass instruments and presenting mainly qualitative data.

Since high pressures had been observed in playing (Papers I-II) and some wide respiratory movements could be expected, as compared to previous applications of RIP, suitability and accuracy were main concerns. One attempt to assess the method was to apply the RIP band around a calibrated cardboard box, which permitted oblique angular distortions, i.e. changes of rhomboid cross-sectional area while keeping constant perimeter, see Figure 15. As shown, the output of the system was linear for a wider range than that equivalent to abdominal dimensions.

Additional calibration procedures were used to check the RIP accuracy in the pressure and lung volume ranges used in wind instrument playing. These experiments are described in Paper IV.

The method was then applied to eight professional players of the four reed instruments, who performed different musical and respiratory tasks. It yielded acceptable accuracy for the measurement of lung volumes, relative abdominal and rib-cage movements and also for the temporal and kinematic details of brief breathing pauses. Particularly challenging demands on respiratory technique were presented in terms of JS Bach's piece (solo obligatto from Cantata 147). Its long, "motoric" phrases played at a fast tempo, did not offer suitable breathing spots to the players. This reflected on the respiratory patterns and also as short duration of the breath pauses. The RIP method turned out to be appropriate even under these extreme conditions. It thus seems applicable to wind instrument performance research in general and should be useful also in pedagogy. In addition, a computer program was written and implemented for the use in portable computers with an acquisition card, allowing real-time calibration and data display.

Breath groups were initiated at 55% - 87% and terminated between 14% - 52% of the players’ vital capacity, depending on instrument, piece, and phrase length. The players generally showed simultaneous and in many cases equally important contributions from rib cage and abdominal wall during playing. These findings often contrasted sharply to the players’ own ideas on how they used their respiratory apparatus. The findings departed from what had previously been found for singers, who tend to resort mainly to chest wall movements for generating lung volume changes. In extreme cases, inhalations were achieved in approximately 300 ms and reasonably synchronised with the RIP signals.

The roles of the authors in this paper were as follows:

LF: Pilot studies on the RIP method, using different calibration procedures and carrying out recordings and data processing from the performance of several musical tasks with the different instruments; design of the experiment; running the experiments; processing of experimental data; writing of the paper; discussion of paper with co-author JS

JS: discussion on experimental protocol; running the experiments; writing of the paper; discussion of paper with co-author LF

Paper V. "Assessment of blowing pressure perception in reed wind instrument players"

Paper V was exploratory in nature since it represented, to our knowledge, the first study ever on this modality of perception in musicians. Paper I showed that professional players change blowing pressure systematically according to instrument, pitch and dynamic level. This indicates that players must possess the ability to accurately sense the pressures they generate. It is obviously a prime control variable along with other parameters, such as the auditory feedback and the sensations in the embouchure.

The investigation applied a psychophysical production method for direct assessment of the blowing pressure perception in professional players. The main object of the study was the investigation of how the players judged mouth pressure, independently of normal playing conditions, i.e. without embouchure effort, reed vibration, airflow or auditory feedback.

The players were asked to produce static mouth pressures corresponding to a set of numbers that were given to them, one by one and in random order. The pressures thus produced were recorded and measured and then compared with these numbers.

The method provided consistent data representing interesting information about this rarely studied modality. Regression analysis revealed that a linear model yielded the highest correlation between measured and perceived pressures. Linearity is not generally found for other perceptual modalities, such as loudness and pitch. The method, however, did not allow for the estimation of the difference threshold, a relevant parameter for quantifying sensation.

The considerable inter-individual differences in sensitivity indicate that average values are not adequate to predict the behaviour of subjects in general terms. We used a limited group of subjects, all of them players of reed woodwinds. It would be worthwhile to include more subjects and players of other wind instruments in future investigation. To complement the result of this paper, other methods could be applied in the future, such as cross-modality comparisons.

Paper VI. "A self-sustained vocal-ventricular phonation mode: acoustical, aerodynamic and glottographic evidences"

Human voice is a result of a complex interaction between anatomical, physiological, neurological and cultural factors. The immense variety of vocal sounds encountered across the cultures of the world demonstrates a multitude of possibilities for using the voice organ (e.g. Zemp, 1996).

The role played by the ventricular folds (Figure 1 in Paper VI) in voice production still constitutes an open issue. Some authors suggested that the ventricular folds were responsible for producing the lowest tones, while the vocal folds would be generating the higher tones. It even has been claimed that the ventricular folds produce falsetto tones, speculations that were refuted already by Garcia (1855). Yet, it has remained unclear how the ventricular folds can be used in sound production, e.g., under pathological conditions such as in ventricular dysphonia (Freud, 1962).

A particular vocal effect, found in certain Asian cultures (Tibet, Mongolia), is associated with a very low fundamental frequency and a dense spectrum. This type of voice production had been documented in previous studies in voice science, ethnomusicology and music acoustics (Smith et al., 1967; Campbell & Greated, 1987; Barnett, 1977; Ellington, 1970), however without exhaustive explanation of the underlying sound production mechanism.

The initial hypothesis was that the mechanism employed includes the ventricular folds. For some years the first author attempted to learn this particular phonatory technique assisted by analysis resources provided at KTH-TMH. The attempts were successful according to expert testimony. This appeared to justify the use of the first author as a subject in the experiment, particularly since Tibetan monks in Sweden seemed hard to access.

Co-operation with the Karolinska Institute, offered the possiblility to obtain images derived from two different techniques, videolaryngo-stroboscopy and high-speed glottography.

The results of Paper VI revealed that, under special conditions, the ventricular folds vibrate periodically in cooperation with the vocal folds. Results did not corroborate the initial hypothesis that the ventricular folds vibrate in a reed-like fashion, i.e., as an inward striking valve. Rather, they vibrated more like the vocal folds and seem to importantly contribute to the primary sound generation in the mode of phonation used in Tibetan chant. The results suggested that, in spite of striking similarities with regard to spectral appearance, this type of voice production is different from the one used in the production of periodical vocal fry, such as in the so-called Strohbass register.

The participation of the authors in the paper was as follows:

LF: idea, planning and design of experiments, serving as the sole subject, data analysis, design of the VVM model, writing of first draft

JS: revision and discussion of the manuscript

BH: providing the possibility to high-speed imaging, expert advice and revision of the manuscript

©1998 by Leonardo Fuks