torsional waves in a bowed string

7/30/2019 Torsional Waves in a Bowed String

1/14

Torsional waves in a bowed string.

This page is an appendix for the scientific paper:Bavu, E., Smith, J. and Wolfe, J. "Torsional waves in a bowed string" (2005, Acustica, 91,

241-246).

It has sound files of the translational and rotational waves, an animationof idealisedHelmholtz motion in translation and a briefsummary of the paper.This figure, reproducedfrom the paper, shows the velocity, in time and frequency domain representation, of the

transverse and torsional waves in a bass E string, bowed by an experienced string player.

For comparison, the torsional wave is represented by r, the product of the string radiusand the angular velocity.

Sound of the transverse wave (2.7 M) (74k)Sound of the torsional wave (2.7 M) (33k)

Notice that both sounds have the clear pitch associated with a highly periodic

signal and that, in both cases, the pitch is that of the transverse fundamental. This pitch

corresponds to the period (26 ms) that is clearly visible in the time domain graphs. Thepitch frequency (40 Hz) is the strongest harmonic in the transverse wave, and is also of

course the spacing between adjacent harmonics. In the torsional wave, it is also the

spacing of harmonics, but the fundamental at this frequency is very weak. The transverse
http://www.phys.unsw.edu.au/jw/reprints/Bavuetal.pdfhttp://www.phys.unsw.edu.au/jw/torsional.html#soundfileshttp://www.phys.unsw.edu.au/jw/torsional.html#animationhttp://www.phys.unsw.edu.au/jw/torsional.html#animationhttp://www.phys.unsw.edu.au/jw/torsional.html#summaryhttp://www.phys.unsw.edu.au/jw/torsional.html#soundfileshttp://www.phys.unsw.edu.au/jw/torsional.html#animationhttp://www.phys.unsw.edu.au/jw/torsional.html#summaryhttp://www.phys.unsw.edu.au/jw/reprints/Bavuetal.pdf


2/14

velocity wave sounds rather like a bowed bass string. The similarity is not surprising:

both are bowed strings. (The spectra are very different, of course. Although the force

exerted on the bridge increases with frequency when compared with velocity, this is inpart offset by the lack of filtering by the radiativity of the instrument.) The angular

velocity wave also sounds somewhat like a bowed bass that has been filtered in an odd

way: the rich harmonic content and the initial transients suggest a bowed string. Theformants around 225 and 450 Hz are near the frequencies of the natural resonances of the

torsional wave, in the absence of a bow. These are so strong, however, that one or both of

the harmonics may be heard individually in the torsional sound file.The animation (made by Heidi Hereth) shows idealised Helmholtz motion of a transverse

wave.

A briefintroduction and

summary

The main

function of aviolin or bass

bow is to induce

a sideways or

transversemotion of the

string. Rosin

placed on a bowensures that

static frictionwith the stringmay be much

greater than

kinetic friction.

Consequently, ina cycle of

normal playing, the string at the position of the bow travels with the bow at a nearly

constant, low velocity in one direction (the stick phase), then slides rapidly past the bowin the opposite direction (the slip phase), as shown in the animation.However the bow

acts on the surface of the string, rather than at its centre, and so also must exert a twisting

or torsional force. This torque excites additional torsional or twisting waves that travel upand down the string. These torsional waves exert only a small torque on the bridge and so

produce little sound by themselves. Nevertheless, they can have an important effect on

the overall sound produced.


3/14

The motion of the point of contact between bow and string

depends on both the transverse speed v of the string, and on

the torsional velocity (its speed is v+r, where r is the

radius of the string). During the stick phase, v+r must

equal the bow speed. The component waves of the familiar

transverse modes of the string are in harmonic ratios and so

produce a periodic wave: one that repeats exactly after oneperiod. However, there is no a priori harmonic relationship

between the torsional and transverse waves. Consequently,

the torsional waves may produce non-periodic motion or

jitter at the bow-string contact. Because the ear is very

sensitive to jitter, this can have a considerable effect on the

perceived sound.

The bowed string has been studied for centuries by scientists, including Helmholtz and Raman. It is

thus a little surprising to discover that the relative magnitudes and phases of the torsional and

transverse motion had not been measured. We did this electromechanically by attaching tiny sensing

coils, using a low bass string to minimise perturbation.

The magnitude of the torsional waves was surprising: they may contribute as much as

tens of percent of the speed at the contact point with the bow, as shown in the figure

above. In the first experiments, the strings were bowed by experienced players. Inmusically acceptable bowing regimes, the torsional motion was always phase-locked to

the transverse waves, producing highly periodic motion. The spectrum of the torsional

motion includes the fundamental and harmonics of the transverse wave, with strongformants at the natural frequencies of the torsional standing waves in the whole string.

Volunteers with no experience on bowed string instruments, on the other hand, often

produced non-periodic motion. This suggests that finding (quickly) the subtlecombination of force and speed that controls the non-harmonic torsional waves is a skill

that string players must learn.

More detail is given in the paper: Torsional waves in a bowed string.

Links with background information

Waves in strings, reflections, standing waves and harmonics. Bows and strings, which includes an animation of Helmholtz motion. An introduction to violin acoustics.

Eric Bavuworked on this project as an undergraduate research project. Other

undergraduates who worked on earlier projects on the bowed string, and who thereforecontributed to this project, are Pierre-Yves PlacaisandManfred Yew.

Harmonic singing (or overtone

singing) vs normal singing

Harmonic singing shares

techniques with diphonicsinging, overtone singing,

xoomi singing, sygyt singing,

throat singing, Tuva singing etc.

We explain some of the
http://www.phys.unsw.edu.au/jw/reprints/Bavuetal.pdfhttp://www.phys.unsw.edu.au/jw/reprints/Bavuetal.pdfhttp://www.phys.unsw.edu.au/jw/strings.htmlhttp://www.phys.unsw.edu.au/jw/strings.htmlhttp://www.phys.unsw.edu.au/jw/Bows.htmlhttp://www.phys.unsw.edu.au/jw/violintro.htmlhttp://www.phys.unsw.edu.au/music/graphics/Eric.jpghttp://www.phys.unsw.edu.au/music/graphics/Eric.jpghttp://www.phys.unsw.edu.au/music/graphics/Pierre-Yves.jpghttp://www.phys.unsw.edu.au/music/graphics/Pierre-Yves.jpghttp://www.phys.unsw.edu.au/music/graphics/Manfred.jpghttp://www.phys.unsw.edu.au/music/graphics/Manfred.jpghttp://www.phys.unsw.edu.au/jw/reprints/Bavuetal.pdfhttp://www.phys.unsw.edu.au/jw/strings.htmlhttp://www.phys.unsw.edu.au/jw/Bows.htmlhttp://www.phys.unsw.edu.au/jw/violintro.htmlhttp://www.phys.unsw.edu.au/music/graphics/Eric.jpghttp://www.phys.unsw.edu.au/music/graphics/Pierre-Yves.jpghttp://www.phys.unsw.edu.au/music/graphics/Manfred.jpg


4/14

acoustics of this style of singing in terms of the measured acoustical response of the vocal

tract. In this technique, the singer emphasises one high harmonic of the voice to such an

extent that it is heard separately from the low pitched note being sung. Different notes inthe harmonic series may be chosen by changing the frequency of the resonance in the

vocal tract that gives rise to it.

For background information on speech and ordinary singing, see ourIntroduction to theacoustics of the vocal tract. For background about our research and techniques, see this

link. On this page, we begin by looking at how the vocal tract behaves for a whisper,

where the resonances of the tract are most clear, then fornormal singing, then forharmonic singing.

Whisper. In the first figure, a subject whispers the vowel in 'hoard'. We show the

frequency response of the vocal tract (For an explanation of the measurements, follow

this link.) The sound of the whisper itself is masked by the injected signal used tomeasure the vocal tract resonances. The figure shows several peaks, indicated by the

arrows. At these frequencies, the sound produced at the vocal folds is most effectively

transmitted as sound produced in the external air. (Technically, these are peaks in the

acoustic impedanceof the vocal tract. At these resonant frequencies, the tract operatesmost effectively as an impedance transformer between the relatively high acoustic

impedance of the tract and the low impedance of the radiation field at the mouth.)

Normal singing. In the figure below, the subject sings the same vowel at the pitch Bb3(117 Hz). In this graph, you can see the harmonics of the voice, and you can see that the

fourth and sixth harmonics appear stronger in the sound spectrum because they are near

resonances of the tract.
http://www.phys.unsw.edu.au/jw/voice.htmlhttp://www.phys.unsw.edu.au/jw/voice.htmlhttp://www.phys.unsw.edu.au/speech/http://www.phys.unsw.edu.au/speech/http://www.phys.unsw.edu.au/jw/xoomi.html#whisperhttp://www.phys.unsw.edu.au/jw/xoomi.html#normalhttp://www.phys.unsw.edu.au/jw/xoomi.html#normalhttp://www.phys.unsw.edu.au/jw/xoomi.html#harmonichttp://www.phys.unsw.edu.au/speech/http://www.phys.unsw.edu.au/jw/z.htmlhttp://www.phys.unsw.edu.au/jw/z.htmlhttp://www.phys.unsw.edu.au/jw/voice.htmlhttp://www.phys.unsw.edu.au/jw/voice.htmlhttp://www.phys.unsw.edu.au/speech/http://www.phys.unsw.edu.au/speech/http://www.phys.unsw.edu.au/jw/xoomi.html#whisperhttp://www.phys.unsw.edu.au/jw/xoomi.html#normalhttp://www.phys.unsw.edu.au/jw/xoomi.html#harmonichttp://www.phys.unsw.edu.au/speech/http://www.phys.unsw.edu.au/jw/z.html


5/14

Over the range shown and for this vowel, this subject's vocal tract has six resonances,which are indicated by the arrows. Note that the subject changes the first two resonances

a little between whispering and singing. The frequencies of these two resonances

determine the vowel in a particular accent. It is not unusual for people to have differentaccents when whispering, speaking and singing. The higher resonances are also

substantially changed, probably because rather different vocal mechanisms are used in

whispering and singing.Harmonic singing. The next graphs show two examples of harmonic singing. In this

technique, one of the vocal tract resonances is made much stronger, while all the others

are weakened. The strong resonance can be made so strong that it selects one of the

harmonics and makes it so much stronger than its neighbours that we can hear it as aseparate note. Hear it is the eighth harmonic that is amplified. Although the fundamental

is only 8 dB lower than the selected harmonic, the fundamental lies in a range in which

our ears are much less sensitive, so it sounds much less loud.


6/14


7/14

For some harmonic singers, more complicated effects than those described here may beinvolved. It has been suggested that, for some sygyt singers, the strong resonance in the

vocal tract may drive an oscillation in the false vocal folds. This could produce a stronger

signal at the high pitch. Further, because the false vocal folds would be nonlinearoscillators, they would produce strong components at integral multiples of the high pitch

frequency, ie at n*f0, 2n*f0, 3n*f0 etc. An example of such a spectrum and an

explanation of the false vocal fold mechanism is given by Chen-Gia Tsai at this link.This research is part of a project investigation the acoustics of singing in general. It is

undertaken byNathalie Henrich,John Smith and Joe Wolfe.

Some related pages and explanatory notes

This style of singing was first popularised in the West by David Hykes, whose

page is at this link. He points out that "harmonic singing" refers to a broader range oftechniques than just the emphasis of an overtone. Chen-Gia Tsai's page on"acoustics of overtone singing" Some interesting results about thetuning of the vocal tract by

Sopranos: resonance tuning and vowel changes
http://www.yogimont.net/jia/overtonesinging/fvfsw.htmlhttp://www.lam.jussieu.fr/Individu/Henrich/http://www.phys.unsw.edu.au/STAFF/ACADEMIC/smith.htmlhttp://www.phys.unsw.edu.au/STAFF/ACADEMIC/smith.htmlhttp://www.phys.unsw.edu.au/STAFF/ACADEMIC/wolfe.htmlhttp://www.myspace.com/davidhykeshttp://www.myspace.com/davidhykeshttp://www.yogimont.net/jia/overtonesinging/http://www.yogimont.net/jia/overtonesinging/http://www.phys.unsw.edu.au/jw/soprane.htmlhttp://www.phys.unsw.edu.au/jw/soprane.htmlhttp://www.yogimont.net/jia/overtonesinging/fvfsw.htmlhttp://www.lam.jussieu.fr/Individu/Henrich/http://www.phys.unsw.edu.au/STAFF/ACADEMIC/smith.htmlhttp://www.phys.unsw.edu.au/STAFF/ACADEMIC/wolfe.htmlhttp://www.myspace.com/davidhykeshttp://www.yogimont.net/jia/overtonesinging/http://www.phys.unsw.edu.au/jw/soprane.html


8/14

In the higher part of their range, sopranos face some

challenges that don't occur in normal speech, and which

don't affect other singers so severely. The problem is that

the harmonics of their voice become so widely spaced

that, unless they make appropriate adjustments, they are

likely to lose the benefit of the resonances of the vocal

tract.The vocal tract usually has two strong resonances in the

range 0-2 kHz. A bass singing a low G has harmonics at

100 Hz, 200 Hz, 300 Hz etc, all the way to several

thousand Hz. He has so many harmonics in this range

that some of them are bound to fall close to the tract

resonances and so he will benefit from enhanced

transmission as sound outside the mouth. A soprano

singing a high C has harmonics at 1000 Hz, 2000 Hz,

3000 Hz up to a similar upper limit. So for a soprano, the

harmonics could 'miss' the resonances for some note-

vowel combinations, and 'hit' for others.

Missing the resonances would be a problem: they are

needed when singing with loud accompaniment, andhaving a resonance near the frequency of the voice can

also make it easier to sing. The distinguished voice

researcher Johan Sundberg hypothised that sopranos

might learn to tune one of their vocal tract resonances to

the pitch of the note they are singing.

This creates a phonetic problem, because the tract

resonances produce bands of increased power

(calledformants) in the voice, and these carryinformation about vowels, information which is mainly

carried in the range 300 to 2000 Hz. For a more detailed

explanation of this point, see Introduction to vocal

tract acoustics. This page provides sound files to demonstrate the effect, background information and a non-technical introduction to the effect.

There is also a soprano challenge. The researchreported here is a spin-off from technologywe

developed for use in language training and speech

therapy.

Abrief scientific reportof the application to sopranosinging is published in the journal Nature and a moredetailed reportis in JASA.


9/14

New results: Wagner makes it easier for sopranos

Wagner is well-known, even notorious, for writing operas that can challenge bothperformers and listeners. Both groups might be surprised to learn that Wagner was

helping both the performers and the listeners by taking the acoustics of the soprano voice

at high pitch into account when he set his text to music. A recent paper from this lab

suggests that this is indeed the case:Smith,. J. and Wolfe, J. (2009) "Vowel-pitch matching in Wagner's operas:

implications for intelligibility and ease of singing", JASA Express Letters, J. Acoust.

Soc. America, 125, EL196. A non-technical account follows.Each vowel in European languages is associated with a set of resonance frequencies of

the vocal tract. For the soprano voice at high pitch, both the intelligibility to listeners and

the ease of production by singers could be improved if the pitch of the note written for avowel corresponded with its usual range of resonance frequencies.

We tested this hypothesis by investigating whether Wagner used certain vowels more

often for the high notes than the low notes, and vice versa. A study of the two great

Wagnerian soprano rles, Brnnhilde and Isolde, indeed found the vowels that required

an open mouth were used more often for the very high notes. Similar studies on someoperas by Mozart, Rossini and Richard Strauss showed no such effect.

We are unaware of any written evidence about Wagner's intentions nor of whether he wasadvised on this issue by sopranos, with whom he sometimes had close relations.

Of course we are not suggesting that Wagner was a better opera composer than others. He

was writing a different type of opera with a much larger orchestra, and making his verydemanding vocal parts somewhat easier.

Summary: It appears that Wagner, either consciously or unconsciously, took the acoustics

of the soprano voice at high pitch into account when setting text he had written to music.This is consistent with the increased importance of textual information in his operas, the

increasing size of his orchestras, and the more complex vocal parts.

More news: resonance tuning in the coloratura/ whistle voice rangeThe resonance tuningdescribed above tunes the first vocal tract resonance to the frequency of the note sung(R1:f0 tuning). It's fine up to about 1 kHz or C6 high C for sopranos. After that, it

becomes difficult to open your mouth any wider. Some sopranos manage for a further

couple of notes, but for many sopranos, the limit of R1:f0 tuning is the limit of their vocalrange.

Coloratura sopranos or the jazz and pop singers who practise the whistle registers,

however, have another technique: starting at around C6, they begin to tune the secondvocal tract resonance to the frequency of the note sung: R2:f0 tuning. Further, they appear

to use a different mechanism and thus to show another transition. This and some related

effects are described in these papers:

Garnier, M., Henrich, N., Smith, J. and Wolfe, J. (2010) "Vocal tract adjustments in thehigh soprano range" J. Acoust. Soc. America. 127, 3771-3780.

Garnier, M., Henrich, N., Crevier-Buchman, L., Vincent, C., Smith, J. and Wolfe, J.

(2012) "Glottal behavior in the high soprano range and the transition to the whistleregisters" J. Acoust. Soc. America. 131, 951-962.

Sound files

These sound files do not form part of the study; they are simply illustrative of
http://www.phys.unsw.edu.au/jw/reprints/Wagner.pdfhttp://www.phys.unsw.edu.au/jw/reprints/Wagner.pdfhttp://www.phys.unsw.edu.au/jw/reprints/highsoprano.pdfhttp://www.phys.unsw.edu.au/jw/reprints/highsoprano.pdfhttp://link.aip.org/link/?JAS/127/3771http://www.phys.unsw.edu.au/jw/reprints/Garnier-JASA-2012.pdfhttp://www.phys.unsw.edu.au/jw/reprints/Garnier-JASA-2012.pdfhttp://link.aip.org/link/?JAS/129/1024http://www.phys.unsw.edu.au/jw/reprints/Wagner.pdfhttp://www.phys.unsw.edu.au/jw/reprints/Wagner.pdfhttp://www.phys.unsw.edu.au/jw/reprints/highsoprano.pdfhttp://www.phys.unsw.edu.au/jw/reprints/highsoprano.pdfhttp://link.aip.org/link/?JAS/127/3771http://www.phys.unsw.edu.au/jw/reprints/Garnier-JASA-2012.pdfhttp://www.phys.unsw.edu.au/jw/reprints/Garnier-JASA-2012.pdfhttp://link.aip.org/link/?JAS/129/1024


10/14

the phonetic effect of the widely spaced harmonics and the resonance tuning. For these

recordings, an experienced soprano, who had no knowledge of the purpose of the study,

was asked to sing an ascending scale over two octaves, from Bb3 to Bb5. She was askedto sing the scale five times, each time using a different vowel sound. The vowels

(phonetic symbols in parentheses) are those in the English words* "hard" (//), "hoard"

(//), "who'd" (//), "heard" (//) and "heed" (//), but sung with an initial consonant "L". (Themusic and the phonemes to be sung were presented in writing.)Scale sung on "La":

Scale sung on "Lore":

Scale sung on "Loo":Scale sung on "Ler":

Scale sung on "Lee":

Let's now listen to the five vowel sounds at low pitch, then at high pitch. (These

files are assembled from the first and last notes on each of the scalesabove.)"La, Lore, Loo, Ler, Lee" at low pitch:

"La, Lore, Loo, Ler, Lee" at high pitch:

Notice that the vowel sounds are much more similar at high pitch+. Do

you think that you can tell them apart? If so, listen to this file, in whichthe order has been changed and see how sure you are.Five vowels at high pitch, order

changed:The difficulty the listener has in differentiating the vowels at high pitch is due to two

effects. One is that high pitch puts the harmonics further apart. (SeeWhat is a sound

spectrum?for details.) The information about which vowel is spoken or sung is (looselyspeaking) conveyed by the relative amplitudes of the harmonics of the voice that fall in

the range from 200 to 2000 Hz. If a soprano sings a high C, there is only one harmonic in

this range, so little vowel information is carried. Another effect is the tuning of the first

vocal tract resonance by sopranos, including this singer. We have studied these effects,which are described very briefly in apaperin Nature, and which are described in much

more detail in a longer paper.We thank soprano Kristen Butchatsky of the School of

Music and Music Education at the University of New South Wales for singing thesamples recorded here and for her help in other aspects of our research.

* English vowels are often presented and studied in the context h[vowel]d, because a set

of such words minimises the number of nonsense syllables. In fact, as soon as we canconvince enough people to call a Head Up Display a 'hud', the list will be complete.

+ Even the consonant is difficult to discern. Much of the information about consonants is

conveyed by the way they modify the vowel that follows (or precedes) them.

Sopranos tune resonances of their vocal tract when they sing in the high range

A non-technical version of reseach published as Joliveau, E., Smith, J. and Wolfe, J. "The

tuning of vocal tract resonances by sopranos", Nature,427, 116.In brief. In the top half oftheir range, but not in the lower half, classically trained sopranos adjust one of theresonant frequencies of the vocal tract to match the pitch of the note they are singing.

This gives greater loudness for given effort, and may have musical advantages, but it

contributes to the difficulty of understanding the words they are singing.

Background.
http://www.phys.unsw.edu.au/jw/sound.spectrum.htmlhttp://www.phys.unsw.edu.au/jw/sound.spectrum.htmlhttp://www.phys.unsw.edu.au/jw/sound.spectrum.htmlhttp://www.phys.unsw.edu.au/jw/reprints/SopranoNat.pdfhttp://www.phys.unsw.edu.au/jw/reprints/Joliveauetal.pdfhttp://music.arts.unsw.edu.au/http://music.arts.unsw.edu.au/http://www.phys.unsw.edu.au/jw/reprints/SopranoNat.pdfhttp://www.phys.unsw.edu.au/jw/reprints/SopranoNat.pdfhttp://www.phys.unsw.edu.au/jw/sound.spectrum.htmlhttp://www.phys.unsw.edu.au/jw/sound.spectrum.htmlhttp://www.phys.unsw.edu.au/jw/reprints/SopranoNat.pdfhttp://www.phys.unsw.edu.au/jw/reprints/Joliveauetal.pdfhttp://music.arts.unsw.edu.au/http://music.arts.unsw.edu.au/http://www.phys.unsw.edu.au/jw/reprints/SopranoNat.pdfhttp://www.phys.unsw.edu.au/jw/reprints/SopranoNat.pdf


11/14

In singing or speech, we can identify two separate effects. First, the vocal folds vibrate

and that produces a sound. The frequency of that sound--the number of vibrations per

second--determines the pitch, ie whether a sound is high or low.Second, the sound fromthe vocal folds passes through the vocal tract. It is modified by the shape of the tract so

that we can make different speech sounds: open your mouth wide and you get 'aah', close

it almost completely and get 'ooo' etc. This process is very familiar to us, but it is also abit subtle. Here's how it works.

The vibration from the vocal folds is a complex sound: we say it has lots of harmonics or

that it is made up of a range of different frequencies. The vocal tract resonates at severaldifferent frequencies and these resonances amplify some of the frequencies present in the

voice. For instance, when this author's mouth and tongue are in a neutral position (when I

say 'er'), it 'amplifies' frequencies of about 450 and 1400 vibrations per second

(approximately the notes A above middle C and the F above the treble clef--both notesbeyond my singing range). Different tract positions amplify different frequency

components of the voice and that allows us to identify different speech sounds. This is

explained in more detail, with diagrams and sound files, in our page Physics in Speech.

These processes are independent. One can keep the vocal tract constant and change thepitch (eg humming, vocalise or singing 'la la la la la'), or one can keep the pitch the same

and vary the vocal tract, which is what the Daleks on Dr Who do ('Ex-ter-min-ate').Normally we do both independently, so we can sing different words on different notes

(normal singing), and we can use different inflexions in the same words. (eg. 'You're not

going' vs 'You're not going?')However, there is a problem for sopranos. In the high range of women's voices*, the pitch

frequency of the notes enters the range of the lowest vocal tract resonance. If the singers

did nothing about this problem, then whenever the pitch of the note coincided with a

resonance in the tract, that note would be much louder than the others. So you would getuneven loudness, and also uneven voice quality. Some time ago, the Swedish acoustician

Johan Sundberg suggested that sopranos actually tune the resonance of their vocal tract to

the note that they are singing: the original evidence for this was that they tend to open themouth more as they sing successively higher notes. However, this could not be confirmed

directly because there was a technical difficulty in measuring the acoustics of the tract

while it was being used for singing.* What about the voices of young children, which are higher in pitch than women's

voices? Children have smaller heads and shorter vocal tracts, so one would expect that

the resonances of their vocal tracts to occur at higher frequencies, so the overlap of pitch

and resonance would occur at higher pitch.

The project on sopranos and resonance tuning
http://www.phys.unsw.edu.au/PHYSICS_!/SPEECH_HELIUM/speech.htmlhttp://www.phys.unsw.edu.au/PHYSICS_!/SPEECH_HELIUM/speech.html


12/14

In the Acoustics Group, we have developed

acoustic techniques for studying the tract during

speech (see Voice Acoustics). Our mainmotivation was to provide technologies for

speech pathology and for language training, but

the techniques are also applicable to singing. Sowe invited in a group of classically trained

sopranos. For these measurements, we produce a

carefully synthesised sound just outside thesinger's or speaker's mouth. A microphone records

not only their voice, but also the way in which

their vocal tract interacts with the synthesised

sound. From the latter, we can tell a lot about theacoustics of the tract. This project became a

research project for Elodie Joliveau, who is both a

physics student and a soprano.What we find with

sopranos is this: In the low range of the voice (below

about A4, but it depends on the vowel), they do just what

we all do in speech and singing: the pitch and the vocal

tract resonances are nearly independent. In the high

range, however, they tune the lowest resonance of the

vocal tract rather precisely to equal the pitch they are

singing. They perform the resonance tuning by gradually

lowering the jaw as they ascend, and/or by 'smiling' more

as they ascend in pitch. What surprised us was the

consistency and precision of the tuning for all vowels.

This resonance tuning gives them uniform loudness and

vocal quality, but it also means that vowel sounds

become very similar--we include above some recordings

that demonstrate this clearly. So sopranos sacrifice someintelligibility in the interests of musical quality.

However, the amount of intelligibility sacrificed is not

great. In the high range, it is very difficult to understand

vowel sounds anyway: because of the high pitch, there is

simply not much frequency information available to the

ear.

This has been remarked on by composers such as

Berlioz, whose book about orchestration warns opera

composers about the effect. Many composers seem to

heed the warning: the high parts in soprano solos

sometimes do not have words, or have a single word

slurred over several notes, or sometimes repeat words

that are heard in other ranges. or sometimes the wordsare simply not important. This is by no means a cricitism

of sopranos: it is just an inherent physical constraint on

the instrument. One doesn't ask a trombonist to play

pizzicato, one shouldn't ask a soprano to make vowel

distinctions in the altissimo. Nevertheless, the effect is

possibly one* of the contributing reasons why opera

houses use surtitles even when the words are in the

language of the audience.


13/14

So, if you are a soprano, download the paper and keep it for defence against conductors who thinkyou should be able to distinguish 'bead' and 'bed' on high C. Or else quote Berlioz (Berlioz, H. Grand

Trait d'Instrumentation et d'Orchestration Modernes (1844; transl. Clarke, M. C., Novello, London,

1882).)

* Other reasons? There are several, starting with the design of the hall. Acoustic

engineers must compromise between a long reverberation time, which makes the soundlouder, and a short one, which makes it clearer. If the performance space was designed for

orchestral music (relatively long reverberation), it will be hard to understand singing.

Then there is the audience, who are not always as silent as may be hoped. (This is often aproblem in the Sydney Opera House: a substantial fraction of each audience is there just

because it is a famous building, not because they like opera.) The orchestra may be too

loud, perhaps because the opera was written centuries ago when string and wind

instruments were less loud than the modern equivalents. Or perhaps because theorchestration is excessive. It must be admitted that not all singers take sufficient care in

pronunciation. And finally, it is especially difficult for the chorus: when plosives (p, d, t,

k etc) are not pronounced in synchrony, it is difficult to discern them.

Soprano challenge

If you are a soprano and you think would like to test whether our observations reflectphysical limitations on all sopranos, or just on some of them, perhaps you would like to

try repeating the exercise recorded in the sound files above. All you need is a microphoneand a computer or tape recorder. (It would help if you had some editing facility such as

the Cool Edit software, but this is not necessary.) First, sing the scale below, senzavibrato, in your professional singing voice, with projection. Depending on your

comfortable range, you might want to make it C major, B major or Bb major.

Then do the same for "Lore", "Loo", "Ler" and "Lee". Then listen to the first notes ineach in each scale. (If you have editing tools, take the first note (the minim or half-note)

of each sample and put them together to make the low pitch file.) Then do the same for

the last note of each scale. Then get a friend to mix up the order of the notes in the finalsample and listen to it. If you can clearly discern them, then we should really like to hear

from you: that would be the basis of a very interesting study!

Reports of the application to soprano singing are published in: Joliveau, E., Smith, J. and Wolfe, J. (2004) "Tuning of vocal tract resonances by

sopranos", Nature, 427, 116. Joliveau, E., Smith, J. and Wolfe, J. (2004) Vocal tract resonances in singing: the

soprano voice, J. Acoust. Soc. America, 116, 2434-39. Garnier, M., Henrich, N., Smith, J. and Wolfe, J. (2010) "Vocal tract adjustments

in the high soprano range" J. Acoust. Soc. America. 127, 3771-3780. Garnier, M., Henrich, N., Crevier-Buchman, L., Vincent, C., Smith, J. and Wolfe,
http://www.phys.unsw.edu.au/jw/reprints/SopranoNat.pdfhttp://www.phys.unsw.edu.au/jw/reprints/SopranoNat.pdfhttp://www.phys.unsw.edu.au/jw/reprints/SopranoNat.pdfhttp://www.phys.unsw.edu.au/jw/reprints/Joliveauetal.pdfhttp://www.phys.unsw.edu.au/jw/reprints/Joliveauetal.pdfhttp://www.phys.unsw.edu.au/jw/reprints/highsoprano.pdfhttp://www.phys.unsw.edu.au/jw/reprints/highsoprano.pdfhttp://link.aip.org/link/?JAS/127/3771http://www.phys.unsw.edu.au/jw/reprints/SopranoNat.pdfhttp://www.phys.unsw.edu.au/jw/reprints/SopranoNat.pdfhttp://www.phys.unsw.edu.au/jw/reprints/SopranoNat.pdfhttp://www.phys.unsw.edu.au/jw/reprints/Joliveauetal.pdfhttp://www.phys.unsw.edu.au/jw/reprints/Joliveauetal.pdfhttp://www.phys.unsw.edu.au/jw/reprints/highsoprano.pdfhttp://www.phys.unsw.edu.au/jw/reprints/highsoprano.pdfhttp://link.aip.org/link/?JAS/127/3771


14/14

J. (2012) "Glottal behavior in the high soprano range and the transition to the whistle

registers" J. Acoust. Soc. America. 131, 951-962.
http://www.phys.unsw.edu.au/jw/reprints/Garnier-JASA-2012.pdfhttp://www.phys.unsw.edu.au/jw/reprints/Garnier-JASA-2012.pdfhttp://link.aip.org/link/?JAS/129/1024http://www.phys.unsw.edu.au/jw/reprints/Garnier-JASA-2012.pdfhttp://www.phys.unsw.edu.au/jw/reprints/Garnier-JASA-2012.pdfhttp://link.aip.org/link/?JAS/129/1024

torsional waves in a bowed string

Documents