Performance Practice
of Electroacoustic Music

Pierre Boulez

Dialogue de l’ombre double


Dialogue de l’ombre double is dedicated to the sixtieth birthday of Luciano Berio and scored for clarinet and tape (pre-recorded clarinet. The piece was composed at IRCAM in 1985, with the technical assistance of Andrew Gerzso (RIM – réalisateur en informatique musicale) and was first performed by Alain Damiens in Florence on October 28, 1985. It takes its title from Claudel’s “Soulier de satin” and it is a fragment of the revised version of Domaines (1968-69). There are two editions of the piece, the “version aux chiffres romaines” and the “version aux chiffres arabes”; they differ only in the disposition of the sections, which are ordered differently. Either way, the transitions remain coupled with their preceding strophes in both scores.

The piece is divided into thirteen sections: a “Sigle initial”, six “Strophes” linked by 5 “transitions”, and a “Sigle final”. The clarinet player on stage plays the strophes, while the “sigle” (initial and final) and the “transitions” are pre-recorded and played back over loudspeakers. The imaginary and diffused presence of the pre-recorded part is thus opposed to the localized physicality of the player. The “clarinette/première” is to be placed ideally in the middle of the hall in order to create a sort of “dialogue” with the pre-recorded tracks (“clarinette/double”), which are played back over a system of loudspeakers made up of six points around the player near the walls of the hall. The audience should ideally sit between the player and the audio system.

The pre-recorded part represents a shadow of the live clarinet, and replies to the musical ideas expressed by the player. The parameter of spatialization plays an important role in the composition, since it is fixed, notated in the score and is to be considered an essential part of the compositional process. According to the instructions in the score, a tape can be rented from the editor as a last option, i.e. in case it is not possible to manage the recording session independently. It is nonetheless strongly advised that the piece be recorded by the same instrumentalist who plays the performance. In order to mix with the electro-acoustical source, the live clarinet should be amplified, if necessary.

Each “Strophe” is centered on a single musical idea, whereas the transitions gradually lead from one section to the other. The dialectic is created diachronically, i.e. the opposition stems from the succession of one part in relation to the next; there is almost no superimposition of parts, which amounts to a sort of textual development. The instruments sometimes mingle «en toute amitié», while in other moments they clash even violently; as a result, a final compromise is found.


Titles of the sections:

Version aux chiffres romains


Sigle initial (Hâtif, chuchoté, mystérieux)

Strophe I (Assez vif, flexible, fluide, calme)

Transition I/II (Flottant, avec des contrastes de vivacité abrupte)

Strophe II (Assez modéré, calme, flottant)

Transition II/III (Très calme)

Strophe III (Très lent)

Transition III/IV (Très lent)

Strophe IV (Très rapide, avec une extrême volubilité, mais ne pas jouer dans la force)

Transition IV/V (Très rapide)

Strophe V (Vif, rigide)

Transition V/VI (Flottant)

Strophe VI (Modéré)

Sigle final (Très rapide, agité, mais murmuré)



Version aux chiffres arabes


Sigle initial (Chuchoté, hâtif, mystérieux)

Strophe I (Très lent)

Transition I/II (Très lent)

Strophe II (Assez vif, flexible, fluide, calme)

Transition II/III (Flottant, avec des contrastes de vivacité abrupte)

Strophe III (Vif, rigide)

Transition III/IV (Flottant)

Strophe IV (Assez modéré, calme, flottant)

Transition IV/V (Très calme)

Strophe V (Modéré)

Transition V/VI (Très rapide)

Strophe VI (Très rapide, avec une extrême volubilité, mais ne pas jouer dans la force)

Sigle final (Très rapide, agité, mais murmuré)


The score (of both versions, “aux chiffres romains” and “aux chiffres arabes”) was edited by Universal Edition, Vienna, in 1985 (UE 18407). It contains an “Introduction”, with detailed information about the structure of the piece, an illustration of the disposition of the loudspeakers along with two diagrams that show two possible dispositions of performer, audience and equipment. Boulez also provides some “Technical Instructions” about the required equipment (audio and lighting) and the amplification of the live part in the appendix. There is also information about procedures to be followed in the live performance and in the recording, including instructions on how to correctly excite the resonance of the piano through the sound of the clarinet. He also gives exact instructions about the spatialization of sound (both manual and automated) and about the lighting. A set of illustrations is provided for this, as well. The tape has to be recorded by the performer, following precise indications stated in the score; if that is not possible, a pre-produced tape can be requested at IRCAM.


Archival material

Paul Sacher Stiftung

Manuscripts: -“Cahier A” from Domains, plan and drafts (with loudspeakers disposition), fair copies, final manuscript, parts, programme notes, etc.-

Audio: Alain Damiens, Ensemble Intercontemporain, P. Boulez, Dir; Paris, Théâtre du Châtelet, 7.10.1989


Other Material

Claude Delangle and Sylvain Malézieu arranged a version of the piece for saxophone. The (unpublished) transcription was edited with a music notation software. According to Mr. Delangle, Boulez himself had suggested the transcription, and the authors testify that he approved of the performance he witnessed. However, the manuscript contains some transposition errors.


Authorial / Editorial instruction


Recording process:

The technical instructions describe in detail how to place microphones and balance direct, indirect and reverberated sound using digital reverb and piano resonances (s. Sound Projection below). During performance, no further transformation of sound is programmed; the audio system is used simply as a means to disseminate the sounds produced by the pre-recorded clarinet.



The score gives two different positions for the live clarinet; in the center of the hall or, if this is not possible, on stage (s. figures 1a and 1b in Technical Instructions, p. 25).


Sound projection

According to the instructions, 6 + 1 loudspeakers are devoted to the spatialization of the pre-recorded part; the other two are used for the amplification of the live clarinet.

As far as routing is concerned, the six loudspeakers are intended to be turning clockwise, starting with #1 downstage right (= left from the audience perspective ). The seventh source is used to create a sound that has to be perceived from the audience as distant (s. score, introduction). A slight amplification of the solo instrument is suggested depending on the hall acoustics. The piano, which is used for reverberation at certain moments indicated in the score should be placed backstage and should not be visible for the audience. An additional loudspeaker will be used to excite the strings with the live signal. The right pedal must be permanently pressed down in order to let the strings resonate. The resonance will be amplified and projected over loudspeakers next to the performer. In Transition 4 to 5 there are two versions of the spatialization. A simpler on (in round shaped cues) and a more complex one intended for automatized performance (in square-shaped cues).



Lighting is used to highlight contrasts between sections of the piece: during the pre-recorded ones the hall must be dark, whereas during the ones played live the centre of the hall has to be illuminated.


Performance report


  1. a) Clarinet ICST 2009, Belix Behringer clarinet / Simon Könz, recording engineer

Studio: ICST

Format: AIFF

SR/bit depth: 44,1 kHz/16 bit

Technical and other characteristics: DAW approach (Logic).


  1. b) Saxophone ICST 2015, Lars Mlekusch saxophone / Florian Bogner, recording engineer

Studio: ICST

Format: Wave

SR/bit depth: 88,2 kHz/24 bit

Technical and other characteristics: Adaptation of the MAX-Patch written by Andrew Gerzon provided by IRCAM.


Report of the recording process at ICST  (transcription for saxophone)


In order to gain more sonic information, we used four microphones instead of two in the recording of the saxophone. No technical specifications of the microphones are given in the instructions. For both direct and indirect signals, microphones with cardioid and a sphere polar patterns were used. The target format after editing was mono. The presence of key clicks as well as the noises generated by the switch between soprano and tenor saxophone were the major difficulties to overcome.

In order to compensate for the dynamic differences of the saxophone (compared to those of the clarinet) a dynamic expander was used; moreover, the study of performances of the original version for clarinet made at IRCAM lead us to assume that the mixing levels, i.e. quantities of reverb, balance, etc., were slightly modified. This freedom of interpretation proved to be necessary in our mixdown process, as well.

Two cardioid condenser microphones and two pick-up microphones were used for recording the response of the piano reverb unit (s. also live piano reverb unit).


Preparation of the Max-Patch (transcription for saxophone)

Three different versions of a performance patch were found in the IRCAM Brahms database (downloaded on 3.9.2015). They are assumed to be variations of a patch written by Andrew Gerzso.

In order to implement the required spatialization, a time list of the cues from the recording for each transition has to be loaded into the patch. This allows the performance of different recordings with own phrasing and time structure.

Sound files can be played from inside the patch or through an external player (e.g. a DAW or a second computer). As far as reverberation is concerned, it is also possible to use a physical model of a piano reverb made with banks of tuned resonators, should a piano not be available, or hiding the instrument from the audience at a given venue not be possible. For practical reasons the patch allows to play back the sound material intended for the seventh loudspeaker throughout the six main speakers.


Sound diffusion (original version for clarinet and transcription for saxophone)

Both the original version for clarinet and a transcription for saxophones were performed at the large hall of the ZHdK (concert halll 3).

– Careful adjustment of start and end volume levels of the tape was necessary in order to have a proper balance with the live played clarinet.

–  In the large hall it was necessary to manually expand the range of the tape towards the lower dynamics in the Transition II à III. A diminuendo until the end of bar 5 to an almost ppp allowed to enhance the slow crescendo towards f present in measure 17. This also applies to the following diminuendo.

–  At the end of the “Sigle final” the perceived dynamic of the closer pp (first clarinet) and the distant ffff (tape) should be of equal intensity. In our performances of both the original and the arrangement, a seventh loudspeaker outside the hall was used.

–  In the large hall, playing back the reverb using only two frontal loudspeakers, as specified in the score, didn’t create a satisfactory sound image. The piano reverberation was therefore routed to all 6 surrounding loudspeakers.


Amplification of the first clarinet / first saxophone

The first clarinet had to be slightly amplified in the large hall in order to avoid an additional attenuation of its timbre. The loudspeakers used for amplification were placed close to the performer. A slight delay was used to provide a clear perception of the position of the instrument. The amplification of the saxophones followed the same approach. Nevertheless, a lower amplification level was required.


Piano reverberation:

Positioning the piano can be a critical issue in venues with no backstage facilities. In this case the piano can be placed in the same room as far away as possible and hidden with screens. A comparison of all three possibilities (piano backstage, piano onstage, electroacoustic resonators) gave the following results:


  1. As expected, the reverberation produced by the physical instrument delivers the best results in terms of sound quality, even though the piano can create additional unwanted sounds and feedbacks. These were avoided by using two pickups and two microphones very close to the sound source and by managing the feedbacks with a sensitive equalization. The microphone signal from the clarinet/saxophone sent to the piano reverb was compressed in order to reduce the perception of the attack. A low threshold, a short attack and a longer release time permit to carve out the resonating tail of the reverberating strings.


  1. Using the piano on stage and isolating it with screens (mainly in order to avoid the perception of the sound of the soloist emerging through the loudspeaker placed under the piano) gives surprisingly satisfactory results in a large hall, comparable to those achieved by positioning the piano on the backstage. Even from the first row (which is 5m away from the instrument), the screens convincingly reduced the aural perception of the direct sound of the clarinet.


  1. The electroacoustic version is very practical for rehearsal purposes, but of course in terms of sound quality not comparable to the two solutions with physical instrument and should only be used in concert if the other two possibilities are for some reason not practical.


A particular case: In the Strophe II (version aux chiffres romaines), at cues n° 2, 4, 6 and 8, two different opinions about the use of reverberation arose. Opening and closing the reverb output produces an unnatural brake of its tail, but at the same time this supports a sharp spatial contrast between “Agité” und “Variable”, similar to what happens in the Transition V-VI. This is in line with the recording instructions, which define “two distinct settings for the levels of DMic plus reverberation. The first, which applies to measures 1-6, 9-11, 14-20, 23-25 and 27 to the end of the Transition, should be such that the clarinet should sound near and natural. The second, which applies to measures 7-8, 12-13, 21-22 and 26 should be such that the clarinet sounds heavily reverberated” (s. score, “Technical instructions”, p. 19). The second approach, opening and closing the reverb input, avoids the unnatural brake and allows smooth transitions.


When choosing the first approach in the transcription for saxophone, both the input and the output fader should be opened and closed together. Leaving the input of the reverb open generates a perceivable staying of the late reflections caused by the spectral density of the alto saxophone, which activates the reverb more vigorously, compared to the B-clarinet.


In our performances of both versions we decided to choose the first, “unnatural” approach. Firstly, the opening gesture was done relatively quickly in order to produce a “wet” resonance. This stop gap worked well in the rehearsal studio, which was rather dry. In the larger hall it had to be performed more slowly than stated in the score. The closing gesture was done as quickly as possible during the breathing caesura.

Selected Bibliography

Barbedette, Sarah, Dialogue de l’ombre double. Lecture d’un titre au kaléidoscope, Bulletin de la Société Paul Claudel, 209 (2013-1), Paris, pp. 53-63.

Colas, Damien, Dialogue de l’ombre double, Traversières magazine, 87/2 (2006), p. 17.

Piencikowski, Robert, Une «musique de la cruauté»? A propos de «Dialogue de l’ombre double» de Pierre Boulez, Figurationen, 5/2 (2004), pp. 47–54.

Ramaut, Béatrice, Dialogue de l’ombre double, de Pierre Boulez: analyse d’un processus citationnel, Analyse musicale, 28 (June 1992), pp. 69-75.

Rogers, Joe, Dialogue de l’ombre double: construction by assemblage, Perspectives of New Music, 38/2 (2000), pp. 30–51.

Tirado, Carmen Noheda, Dialogue de L’ombre double, “self-paráfrasis” espacializada”, Quodlibet, 57/3 (2014), pp. 28-58.

Vidolin, Alvise; Bonsi, Davide; Diego, Gonzalez; Stanzial, Domenico, Convolving a clarinet with a piano, Forum Acusticum, (2005), Budapest.



Andrew Gerzso, Analyse de Dialogue de l’ombre double de Pierre Boulez:


CD releases (selection)

-Aeon (2008), AECD 0860, Vincent David (sax) (Rec. Paris, IRCAM, Espace de projection, 2007).

-Deutsche Grammophon (1998), DGG 457 6052 – 20/21 Series, Alain Damiens (cl.), Andrew Gerzso.^–

-Erato (1995), Erato 9984952, A. Damiens cl., A. Gerszo mus. Assistant (Rec. Paris, IRCAM, 03-09/1990).

Schematic Overview
Composer Pierre Boulez
Title Dialogue de l’ombre double
Year 1982-85
Studio IRCAM
Type Clarinet and tape
Duration ca. 18’
Publisher Universal Edition Vienna


Rent material:

Renter Universal Edition Vienna, n° UE 18407
Format Score


Archival material:

Paul Sacher Stiftung Drafts and sketches; Audio material


Commercial releases (selection):

Deutsche Grammophon (1998), DGG 457 6052 – 20/21 Series, Alain Damiens (cl.), Andrew Gerzso
AEON (2008), AECD 0860, Vincent David (sax), (Rec. Paris, IRCAM, Espace de projection, 2007)
Erato (1995), Erato 9984952, A. Damiens cl., A. Gerszo mus. assistant, P. Boulez cond. (03-09/1990, Paris, IRCAM).