The term “overtone singing” refers to a singular vocal technique in which a single person sings in two voices: a drone formed by the fundamental sound, and a superimposed melody formed by harmonics.

This article is the result of two complementary approaches: a pragmatic research through the learning and exercise of overtone singing that Trân Quang Hai has been carrying out since 1971, and a visualization research conducted on the physiological and acoustic level for the preparation and production of the film Le chant des harmoniques, directed by Hugo Zemp in 1988-89. In this film, Trân Quang Hai is the main actor, alternately singer and ethnomusicologist: teaching overtone singing during a workshop, interviewing Mongolian singers, lending himself to radiocinematography with computer image processing, and singing into the spectrograph’s microphone to then analyze his own vocal technique (2). The spectrographic images that we discovered practically at the same time as we were filming them – the latest Sona-Graph model allowing the analysis of the sound spectrum in real time and it synchronously had arrived at the Department of Ethnomusicology of the Musée de l’Homme a few days before shooting – encouraged us to pursue these investigations and to lead them in a direction that we would probably not have envisaged without the making of the film (3).

2 The details of the making of this film – which premiered on July 27, 1989 at the Congress of the International Council for Traditional Music in Schladming (Austria) – are described elsewhere (cf. Zemp 1989). A booklet, to accompany the edition in the form of a videocassette, is currently in preparation.

3 Research carried out in close collaboration by the two co-authors who each made – in addition to the joint evaluation of each stage of the work – specific contributions. The sonograms, as well as the detailed analyzes used to write the legends of the figures, are from Trân Quang Hai who, moreover, is at the same time “privileged informant” and singer of 26 recordings reproduced on sonograms. The research design, formatting and writing of the article is by Hugo Zemp.

The use of spectrographic tools for the analysis of overtone songs is not new: Leipp (1971), Hamayon (1973), Walcott (1974), Borel-Maisonny and Castellengo (1976), Trân Quang Hai and Guillou (1980 ), Gunji (1980), Harvilahti (1983), Harvilahti and Kaskinen (1983), Desjacques (1988), Léothaud (1989). There is no question, within the framework of this article, of evaluating this work, of summarizing its results or of giving its history. In the most recent study, G. Léothaud (1989: 20-21) (4) summarizes excellently what he calls the “acoustic genesis of overtone singing”.

4 Appeared in a new publication: Dossier nº 1 of the Institut de la Voix, Limoges. In addition to two brief reports relating to clinical and paraclinical examinations of the phonatory apparatus and the overtone emission of Trân Quang Hai, examinations carried out one by an O.R.L. (Sauvage 1989) and the other by a phoniatrist (Pailler 1989), and an extract from a presentation on the realization of overtone singing (Trân Quang Hai 1989), this file also contains the most complete bibliography and discography to this day concerning overtone singing.

The phonatory apparatus, like any musical instrument, is made up of an exciting system, here the larynx, and a vibrating body responsible for transforming the energy    received into acoustic radiation, the pharyngo-buccal duct.

The larynx delivers a harmonic spectrum, the primary laryngeal sound, determined in frequency, of homogeneous appearance, that is to say devoid of notable formants – therefore of vocalic color – and whose richness in harmonics varies essentially according to the vibratory structure of the vocal cords […].

This primary supply passes through the pharyngo-buccal cavities, undergoing significant distortions there: the pharynx and the mouth therefore behave like Helmholtz resonators, and this for all frequencies whose wavelength is greater than the largest dimension of these cavities. […].

The parameters determining the natural frequency of the phonatory cavities can vary in considerable proportions thanks to the articulator system, in particular by the mobility of the jaw, the opening of the mouth and the position of the tongue. This, above all, can divide the oral cavity into two resonators of smaller volume, therefore of higher natural frequency. In other words, the oral cavities can continue to behave as Helmholtz resonators even for very high harmonics of the laryngeal spectrum, those whose wavelength is small, and in any case less than the length of the pharyngo-pharyngeal duct. oral.

Overtone emission consists for the singer of emitting a spectrum rich in harmonics, then very finely tuning a phonatory cavity to one of the components of this spectrum, the amplitude of which thus increases sharply by resonance; by moving the tongue, the oral volume can vary, therefore the natural frequency, and in this way select different harmonics.

It proposes an analysis grid, centered on four levels and twelve relevant criteria. 1º Characteristics of the voice spectrum; 2º Nature of the overtone formant; 3º Characteristics of the melody of harmonics; 4º Field of freedom of diphonic fluctuation. The application of this grid makes it possible to deepen and systematize the spectral analysis of overtone singing which can now be based on many new sound documents recently published on discs, adding to the well-known old ones. However, that is not our goal.

We propose to examine how the different styles or stylistic variants of overtone singing – called among the Mongols khöömii (5) (“pharynx, throat”) and among the Tuva of the USSR khomei (from the Mongolian term) – are physiologically produced . In this field, the descriptions are rare and not very detailed, whereas we have known for many years the vernacular names designating these styles among the Tuva, whom Aksenov (1973: 12) thinks form the center of the Turco-Mongolian culture. overtone singing, since they practice not just one but four stylistic variants (kargiraa, borbannadir, sigit, ezengileer; a fifth name, khomei which is at the same time the generic name of overtone singing, replacing in some places the term borbannadir) . Implicitly, the neighboring peoples he cites – Mongols, Oirats, Kharkass, Gorno-Altais and Bashkirs – would know only one style. In any case, for the Mongols and for the Altai of the USSR, mountain dwellers living in the chain of the same name, this is not correct. The latter use three styles named on the notice of a record sibiski, karkira, kiomioi (Petrov and Tikhonurov). The most famous overtone singer in Mongolia and abroad, D. Sundui, listed five styles during the Musical Voices of Asia festival in Japan: xarkiraa xöömij (narrative xöömij), xamrijn xöömij (nose xöömij), bagalzuurijn xöömij ( throat xöömij), tseedznii xöömij (chest xöömij), kevliin xöömij (belly xöömij), the last two being generally not differentiated (Emmert and Minegishi 1980: 48). In the interview for the film Le chant des harmoniques, T. Ganbold indicates the same five names. He briefly introduces the first four styles, adding that he does not know how to do “belly khöömii”, thereby distinguishing it from “breast khöömii”. But he does not explain how he produces these different styles. It is true that the interview had to be done in a very short time, saving film, and with the help of a translator, an official of the Ministry of Foreign Affairs of Mongolia, probably unfamiliar with the subtleties of singing. . T. Ganbold and G. Iavgaan had also led several workshops at the Maison des cultures du monde in Paris; this time the translation was provided by an ethnomusicologist, Alain Desjacques, but the two singers were not more explicit. As for D. Sundui, to whom a Japanese musicologist asked how to learn overtone singing, he replied simply that you had to know how to hold your breath as long as possible, use it effectively, then listen to sound recordings and try (Emmert and Minegushi 1980: 49).

5 The transliteration changes according to the authors xöömij, khöömii, chöömij, ho-mi.

Despite the fact that his country of origin (Vietnam) and his host country (France) do not traditionally know overtone singing – or perhaps because of it – Trân Quang Hai succeeds in reproducing different styles or stylistic variants , or at least come close. Having learned without receiving instructions or advice from experienced singers, and without being able to rely on published descriptions, he was forced to proceed by trial and error. This empirical, but nevertheless systematic research, allowed him to become aware of what is happening at the level of the oral cavity. Conducting introductory workshops on overtone singing for many years has led him to know how to explain it.

The originality of the new research presented here consists of 3 points:

1. Trân Quang Hai tries to imitate as best as possible the songs reproduced on the sound recordings we have. For this, he relies on both auditory and visual perception, trying to obtain, on the Sona-Graph monitor, tracings of spectra similar to those of singers from Mongolia, Siberia, Rajasthan and from South Africa.
2. He subjectively describes what he does and feels physiologically when he gets these tracings.
3. In order to better understand the mechanism of the different styles and to explore all their possibilities – even if they are not exploited in traditional overtone singing – he performs experiments that no one has probably ever tried.

This research could not be carried out with old-fashioned spectrographs used until 1989 by the authors mentioned above. This required a device capable of restoring the sound spectrum in real time and synchronous sound, the DSP Sona-Graph Model 5500 that our research team acquired in December 1988. If we change, while singing, the parameters of the vocal emission, we immediately see the layout of the harmonics change. Thanks to the feedback of the new tracing, the voice output can again be changed. Thus, the research is strictly experimental. In the first study on the acoustics of overtone singing, E. Leipp reproduces a theoretical diagram of the phonatory apparatus, showing five main cavities as resonators: 1º the pharyngeal cavity; 2º the posterior oral cavity; 3º the anterior oral cavity, the tip of the tongue directed towards the palate separating cavities 2 and 3; 4º the cavity between the teeth and the lips; 5º the nasal cavity (Leipp 1971). The exact role of these different cavities seems difficult to define.

Thanks to his pragmatic experience as a singer and pedagogue, Trân Quang Hai was led to distinguish two basic techniques essentially using one oral cavity or two oral cavities (Trân and Guillou 1980: 171), the two techniques being able to be more or less nasalized . In the one cavity technique, the tip of the tongue stays down, like when pronouncing vowels. Trân Quang Hai found this technique perfect to make beginners feel better the modification of the buccal volume with the pronunciation of vowels. He tells the trainees that they must “leave the tongue in a resting position” (cf. the film Le chant des harmoniques). The X-ray images from the film, however, show that the back of the tongue rises during the successive pronunciation of the vowels o, ɔ, a (this is not related to overtone singing). Radiologist F. Besse speaks of “the ascent of the tongue”. The “tongue rest” metaphor still holds true in the sense that the tip of the tongue stays down. The radiological image shows that in this technique there is contact between the soft palate and the posterior part of the tongue, separating the oral cavity from the pharyngeal area.

In the two-cavity technique, the tip of the tongue is applied against the roof of the palate, thus dividing the oral volume into an anterior cavity and a posterior cavity. Here there is no contact between the back of the tongue and the soft palate; the posterior buccal cavity and the pharyngeal cavity being connected by a wide passage. The selection of different harmonics to create a melody can be done in two ways: a) the tip of the tongue moves from back to front, with the highest harmonic being obtained in the most forward position; the anterior oral cavity is then reduced to the maximum (cf. the radiological images of the film Le chant des harmoniques); b) the tip of the tongue remains glued to the palate without moving, the harmonics being selected according to the greater or lesser opening of the lips: from the smallest opening when pronouncing the vowel o (Low harmonic) until ‘at the widest opening when pronouncing the vowel i (high harmonic). This second way does not seem to be used by the Mongolian singers that we have been able to observe, and Trân Quang Hai only uses it for its pedagogical interest (comparison with the one-cavity technique) during his initiation workshops.

In order to explore all the possibilities of the two main techniques, Hai sang harmonic scales from different pitches of the fundamental. For the one-cavity technique, we see in Figure 1 that the harmonics that can be used to create a melody only slightly exceed the upper limit of 1000 Hz, whatever the fundamental. But the lower the fundamental, the more harmonics there are. Thus, for the lowest fundamental (90 Hz, approximately one fa1) of fig. 1, usable harmonics are H4 (360 Hz), 5, 6, 7, 8, 9, 10, 11, 12 (1080 Hz), resulting in the scale (transposed) C, E, G, B♭ – , do, re, mi, fa#-, sol. For the highest fundamental (180 Hz) of fig. 1, only harmonics 3, 4, 5 and 6 are exploitable, and the resulting scale, sol, do, mi, sol, is much poorer in melodic possibilities.

With the two-cavity technique (fig. 2), and the lowest fundamental (110 Hz = A1), Trân Quang Hai manages to bring out the harmonics between H6 (660 Hz) and H20 (2200 Hz). To create a melody in the highest zone, it is necessary to select even or odd harmonics (see below fig. 11 and 12), since the harmonics are too close together for a musical scale. The transmission of the highest fundamental (220 Hz) of fig. 2 selects from H4 (880 Hz) to H10 (2200 Hz).

A quick glance shows that, in fact, the harmonics obtained by the one-cavity technique are essentially located in a zone up to 1 KHz, whereas the harmonics obtained by the two-cavity technique are placed mainly in the area from 1 to 2 KHz.

Traditionally, Mongol and Tuva women did not practice overtone singing. According to the singer D. Sundui, this practice would require too much force, but there would be no prohibition on this subject among the Mongols (Emmert and Minegushi 1980: 48). Among the Tuva of the Soviet Union, overtone singing is said to be almost exclusively reserved for men; a taboo based on the belief that it would cause infertility to the woman who practiced it would gradually be abandoned, and some young girls would now learn of it (Alekseev, Kirgiz and Levin 1990). These authors further say that “women are able to produce the same sounds, albeit at higher pitches”, which is only partially true. It’s true if we talk about the “sounds” of the drone which are higher for a female voice than for a male voice, but it’s not true for the melody of harmonics which cannot go higher than in men. This can already be deduced by examining Figs. 1 and 2 where the upper limit of harmonics obtained from the highest fundamental (180 and 220 Hz) is not higher than the upper limit of harmonics obtained from the lowest fundamental, one octave lower (90 and 110Hz). Confirmation can be found by examining Figs. 3 and 4, reproducing the voice of Minh-Tâm, the daughter of Trân Quang Hai (6). With the technique with one cavity and a fundamental of 240 Hz, the number of harmonics is very limited H3 to H5 (1200 Hz). With the two-cavity technique and a 270 Hz fundamental, harmonics 4 (1080 Hz) to 8 (2160 Hz) can be used to create a melody, resulting in a richer scale (transposed C, E, G, B ♭-, C). It follows that a high-pitched woman’s voice does not allow melodies to be created using the single-cavity technique. The Xhosa woman from South Africa recorded by R.P. Dargie, who however uses this technique (as shown in Figs. 7 and 8), has a deep voice, in the register of male voices (100 and 110 Hz = G1 and A1).

6 She was seventeen when this recording was made, but her father had taught her overtone singing from the age of six.

If the conclusions that we have drawn from these experiments (fig. 1 to 4) are correct – and we believe that they are – we should be able to deduce that the styles of overtone singing whose sonograms present a melody of harmonics do not not exceeding essentially 1 KHz are obtained according to the one-cavity technique, while those whose melody of harmonics lies essentially between 1 and 2 KHz are obtained according to the two-cavity technique. The experiments made by Trân Quang Hai, trying to imitate the different stylistic variants, confirm this. In the lines that follow, we will examine the physiological characteristics of the different stylistic variants of overtone singing, identifying three criteria: the resonator(s); muscle contractions; ornamental processes.