un coup de dès

20 July 2007 - musik muzikologi


mémoire de composition pour le cours d’électroacoustique, à partir de Un coup de dès de Mallarmé, sous la direction du Dr. Ian Stonehouse, Goldsmiths College – a obtenu un First – en anglais

Option 3-text sound composition:
A studio composition in which the voice is the only or key sound source. You should select your text(s) from a daily newspaper. You may include other sounds, however human utterance derived from readings of the text should drive the work.


I have chosen to use Mallarmé’s Un coup de dès , “an imaginative exercise in statement and parenthetical commentary that can be read in different ways and outwardly resembles the kind of learned analysis of a haiku in a literary magazine that smothers the original text in editorial overgrowth.” In addition, “the essential meaning of the poem does not change in whatever order a reader chooses to read the text, parentheses and commentaries”. This text has therefore two advantages: a multiplicity of the possible reading paths, and a graphical layout which lends itself particularly well to a transposition in the sonic domain, allowing both the visual diachronic and synchronic dimensions to be preserved and/or augmented. It is therefore my own personal reading path of Un coup de dès at this particular moment in time that I am illustrating. Further, I have fully integrated in my approach the element of hasard or chance, which is the poem’s main theme, as reflected in its title.

production of initial material
In agreement with the tutor for this course, a throw of a dice dictated that I should use p. 2 and p. 5 of Mallarmé’s poem , as using all six pages would not have been feasible in the context of this assignment. To facilitate the recording of the sentences, I isolated twenty-eight blocks of sentences . This was done on a visual basis, and therefore has no structural significance. I then recorded each of the sentence blocks, using a handheld recording device (Zoom H4, set at 44.1Khz, 16 bit), using my own voice as well as the voices of strangers, by going around college and asking people at random, in random parts of the college. No reading instructions were given to any of the readers. I consciously discarded the option of recording the voices in the college’s studio so as to avoid clean and near perfect results, which, on aesthetics grounds, I consider to be an inappropriate narrative overlay. Each recording was used “as is”, with no correction or retake, my operative motto being that once a dice is thrown, then it is thrown, it cannot be “unthrown” if the result doesn’t suit your purposes. It was therefore important that this chance element be transposed in the recording process, by accepting the conditions of the recording’s moment, such as ambient noises or hesitations in reading. In a way, I recorded the voices as I found them.

overall structure
The number SIX (the six sides of a dice) was taken as an operative number. The work therefore consists of six different clusters of all twenty-eight sentences, framed by an Introduction (at the start of the piece) and an Epilogue (and the end of the piece). Each sentence is assigned to one of six tracks and with its own particular pan, and each track is sent to a Buss which has it own plugin FX as an insert. The busses are then routed back into Pro Tools main mix. Throughout, there is a sonic bed, which consists of repetitions of Phrase 00_B (on Track 8 with no FX), which is Phrase 00 treated in Max-MSP, using a freeware granulation patch called Granular 2.0, with Preset 7, of my own making (Cf. Figure 3-Max-MSP setting for Preset 7).

This consists of Phrase 00_A (on Track 7 with no FX), itself being Phrase 00 processed in Max-MSP, using a freeware granulation patch called Granular 2.0, with Preset 1, of my own making (Cf. Figure 1-Max-MSP settings for Preset 1).

Each cluster was obtained by a random shuffling procedure of all 28 sentences, each sentence being assigned at random to one of six tracks, and given its own unique pan value in the range of [-64, +64], also obtained at random. All the assignment calculations were done in Mathematica V5.2 (Cf. Figure 5-Mathematica Notebook-part 1 of 2 for the algorithms). In each cluster, each sentence therefore appears only once, but in a different order (Cf. on p13 & 14 for a list of all six clusters and their parameters).

This consists of six extra sentences, all heard at the same time. They were produced by mixing down (with no FX) each cluster separately, then processing them, also separately, in Max-MSP, with Preset 6, of my own making (Cf. Figure 2-Max-MSP settings for Preset 6). Each cluster processing in Max-MSP was recorded for 45 to 60 seconds on the DAT recorder and transferred back into Pro Tools.

Pro Tools 6.4 LE implementation
Within each cluster, each sentence was positioned manually so as to reflect, more of less, the printed layout of Un Coup de dès. Just before the start of each cluster, the words “un coup de dès” , extracted from Phrase 00_A, can be heard, as a kind of punctuation.

Each of the main six tracks is sent to a unique buss (via Send a), each of which has a different FX plugin as an insert. The amount of Send is fixed throughout. Each buss is then sent back into Pro Tools main path on its own stereo track. Each of these Return tracks has its own varying volume curve, giving a variable ratio of Dry to FX throughout, so as to give “movement” to the mix, while all of the plugins Dry outputs are switched off (where applicable). The Volume curves are drawn manually, and have no structural significance.

The choice of FX plugins was guided by my own recording studio experience. I settled on 5 of the GRM Tools Classic VST plugins pack , and The Waves Trueverb plugin, as available in the College’s studio at the time of this assignment. I have found the GRM plugins to be of excellent quality and conception. In particular, the capability to morph from the parameters of one preset to another, on a time scale of 0 to 20 seconds is ingenious. In addition, Preset 15 randomises the current settings to +/- 10% of their values, while Preset 16 generates a complete new random set of values. This is equivalent to throwing the dice. As for the Trueverb plugin, I consider it as the best reverb plugin currently available, because of its very smooth tail, and have been using it in all my productions since version 2.8. The preset’s setting for this plugin is of my own making, and is fixed throughout the piece. The assignment of plugins to busses is not structurally significant.

For the Epilogue, each of six new sentences were placed on their own Stereo track, with no FX, and given their own starting Pan values. Over the duration of each sentence, its Pan was made to move gradually to its opposite value (e.g. new sentence 1 has its Pan varying from -49 to +49).
The final mix was done manually (not bounced), manually generating in turn for each of the five GRM plugins a random set of values (Preset 16), so as to give a dynamic result. That is Pro Tools was put in Play mode, and the output recorded on DAT. Only one take was done, so as to let chance, again, be fully operative.

Pro Tools routing
Track 1 is sent to Buss 1-2, which has the GRM Tools Classic VST Comb Filter plugin (adds power and resonance to a sound, transform a sound’s timbre, smooth rhythmic sounds into long continuous sounds) as an insert.
Track 2 is sent to Buss 3-4, which has the GRM Tools Classic VST Pitch Accum plugin (creates two distinct ‘shadows’ of a sound at different transposition levels and at different delay intervals) as an insert.
Track 3 is sent to Buss 5-6, which has the GRM Tools Classic VST Reson plugin (creates new sounds from current sounds by rebalancing and redistributing the sound’s resonant frequencies) as an insert.
Track 4 is sent to Buss 7-8, which has the GRM Tools Classic VST Shuffling plugin (creates an unusual resonance or reverberation, fill an audio space with overlapping fragments of a sound, turn a single voice into a crowd) as an insert.
Track 5 is sent to Buss 9-10, which has the GRM Tools Classic VST Doppler plugin (lets you move sounds through an audio space with changes in pitch that corresponds to their movements) as an insert.
Track 6 is sent to Buss 11-12, which has the Waves Trueverb plugin (a reverb plugin) as an insert.

My procedure has used a combination of found material (stranger’s voices) and chance throughout the manipulation of the material. It therefore constitutes a template which can be used many times without giving similar results. This includes the Live Mixing aspect, which could be re-created for public performances, in particular on a multi-speaker set up.

A cause de spam trop nombreux, les commentaires sont fermés-envoyez tout commentaire par mail à bdeflorence@gmail.com ou bien apportez votre parole au groupe de discussion the-letter sur google group.