The Genealogy of Transfigured Wind
II. Form and Process
The proposed form of the whole composition — where the seminal flute solos would be situated in the large design — as well as the digital signal processing strategies that were conceive and realize, is described below through a series of page images and sound files. This abundance of sound examples should allow you to experience, even now at a remove of more than two decades, some of the sense of discovery we had then.
Structure for Transfigured Wind II (III)
- Figure 31 - An early diagram showing the proportionally expanding durations of Solo flute materials (top line, in four divisions) and those of the Ensemble (bottom line, in four divisions), and how they are to interact in formal alternation. Note that the Ensemble materials are further stratified into proportionally expanded strata for woodwinds, percussion, brass, and strings.
- Figure 32 - Fully detailed overall plan for Transfigured Wind including proposals for the structure of the Tape cues.
Preliminary Processing Tests
Audio: Various computer transformations were attempted on the Harvey Sollberger preliminary recordings (primarily on the opening phrase of Debussy's Syrinx). Here are some of the results achieved before the composed solo materials were recorded in October of 1983.
-
30 - The subject phrase from Syrinx is transposed very slightly so that the original and the altered copy are de-tuned. The stereo presentation allows differing groups of partial components to be heard as well as the combination.
Play: Left Channel | Right Channel | Stereo (0:05.5) - Note: There is an echo effect. -
31 - The same subject pair under more substantial de-tuning. Differing groups of partial components can be heard on each channel.
Play: Left Channel | Right Channel | Stereo (0:05.5) -
32 - Progressive detuning for a very marked effect. Differing groups of partial components can be heard on each channel.
Play: Left Channel | Right Channel | Stereo (0:05.5) -
33 - The original subject phrase with added vocal-like vibrato (amplitude modulation superimposed at a rate of 180 msec. periods). Differing groups of partial components can be heard on each channel.
Play: Left Channel | Right Channel | Stereo (0:05.5) - 34 - The Syrinx subject now with superimposed accents achieved with the Karplus-Strong, plucked-string algorithm (with an 80 msec. period) and reverberation. (0:06)
- 35 - This transformation (34) becomes a source for comb filtering. (0:11)
- 36 - The original subject comb filtered without the Karplus-Strong addition. (0:11)
- 37 - The 10 component pitches of the Syrinx phrase are heard, each separated by an interval of time. (0:09.5)
-
38 (- 47) - The 10 Individual notes from Syrinx phrase.
Play: sb05 | sb07 | sb09 | sb11 | sb13 | sb15 | sb17 | sb19 | sb21 | sb22 - 48 - The 10 constituent notes joined again, but in reversed order. (0:05)
- 49 (50) - The constituent tones of the Syrinx phrase played in random order but with an increasing probability of shorter durations over time. These are simulations of the SPIRLZ algorithm's effect, but without its logical rigor. Spatialization on an elliptical path in a reverberant virtual environment adds richness. Play: Example 1 | Example 2 (14.5")
Processing Tactics
- Figure 33 - First notes on digital processing tactics: 9 categories. [Notebook]
- Figure 34 - An elaborated list, now of 12 tactics, each with a 6-letter designation (e.g., EXTEND, SPIRLZ, GOSTNG, etc.). They are arranged below in four groups of three. [Notebook]
- Figures 35 (- 39) - Each of the 12 tactics is explained with a diagrammatic representation. [Notebook]
Now, given both the recorded versions of the four Solos and specific processing approaches, notes were made on the possible signal processing treatment of Solo materials for each of the four Tape Cues, as well as where in the various formal strata each of these processed elements would fit:
-
Figures 40 (- 50) - The Transfigured Wind solo score with bracketed source identities:
- source "s1a" is the initial sffz, used in phase vocoded extension in 51 & 52 (see below)
- source "s1f" is the final phrase in Solo I, used in SPIRLZ transformations in Tape I: 56 & 57 (see below)
-
Figure 51 (52) - Notes on proposed processing for Tape I.
Note here, for example, the elongation proposed for the s1a sf, and, in 53, the SPIRLZ processing of the final s1f phrase in Solo I. Both processes were a part of the final Tape I cue. - Figure 53 (- 55) - Notes on proposed processing for Tape II.
- Figure 56 (- 58) - Notes on proposed processing for Tape III.
- Figure 59 (- 61) - Notes on proposed processing for Tape IV.
Instances of Processing Tactics
All work on Transfigured Wind at CME's CARL facility took place in a computer software environment designed and realized by F. R. Moore. Cmusic was an extension of work done at Bell Telephone Laboratories, beginning in the early 60s. Moore's extension allowed wider utility and dissemination of the potential for digital synthesis through the then emerging mid-sized computers such as UCSD's PDP 11-55 DEC machine.
Audio: In the following examples, the initial item designates the subject sound, while the sound files following each present preliminary transformations of various kinds.
EXTEND (pvoc)
flute sf (sf1):
- 51 - Original sf impulse with 2 phase vocoded expansions. (0:15)
- 52 - Phase vocoded extension x 128. (0:29)
flute multiphonic with simultaneous rising hum (s4h):
- 53 - Rising 3-note phrase from Solo IV (0:09)
-
54 - A composite of four individual phase vocoded expansions by a factor of 9
Play: Left Channel | Right Channel | Stereo (1:23.5)
- Figure 62 - A cmusic music computer "score" for phase vocoding (pvoc) processing , with comments. This structure of text lines (minus the parenthetical comments provided here in what was termed the "verbose" text file mode) is the sort of designation -- the score -- that one prepared in order to direct the computer to apply a particular transformation to a digitally represented subject sound file.
- Figure 63 - A temporal design sketch for extending a seminal 8-note phrase (s4a) from the beginning of Solo IV. This stretch initiates Tape IVb. Each pitch in the subject phrase was extended by phase vocoding to the proper duration to achieve the indicated "augmentation" in time. Smaller numbers are times within the local phrase, larger numbers time, in seconds, within the overall design of the entire work. (The computer cues must, of course, be planned to coordinate productively with what the composed instrumental music is doing at the same time.) [Notebook]
- Figure 64 - Diagrammatic exploration of the coordination between the elongated pitches and other, simultaneously occurring materials. [Notebook]
SPIRLZ
Solo I content (final phrase):
- 55 - This is the subject phrase (s1f) from the end of Solo I. (0:05)
-
56 - A seven-cycle SPIRLZ algorithmic treatment of s1f. (1:31.5)
The SPIRLZ algorithm fragments an original sound file by, in effect, "spiraling out" from a chosen center point, extracting segments of a specified size in alternation: first later than the center point, then earlier, etc. This process repeats for a certain number of "cycles". Producing a gradually accelerating intensity in the output. - 57 - A four-channel composite of varied SPIRLZ treatments of the same subject, superimposed. (0:29)
Solo IV content (trill passage):
- 58 - A complex trill passage (s4i) from Solo IV. (0:10.5)
-
59 - An extension (approximately 5 x) of s4i over several cycles. (0:55.5)
At first the basic trill speed remains constant as the algorithm extracts intact chunks, but, by the end, the size of the extracted elements is smaller that the individual trill notes, so intensity has been incrementally raised.
Slow harmonics oscillation:
-
60 - A set of four superimposed SPIRLZ processing of a slurred harmonic source. (0:46)
Note that, while the algorithmic process is identical, the "feel" of the product is totally different.
SPLITZ
Cadential phrase within Solo III:
- 61 - Fragment s3d11 from the Solo III recording. (0:06)
-
62 - The fragment s3d11 expanded by a factor of 5 through application of the SPLITZ algorithm in two nested passes. (0:28)
This algorithm segments the subject soundfile into a specified number of proportional fragments. The odd numbered fragments are spaced out increasingly in the output result. The even fragments are presented in reversed order and become, over time more closely spaced. The effect here is a "stuttered" rhythm created by overlapping series of fragments that diverge or converge in time.
SMEARM/SCATER
Composite source from several Solos:
- 63 - The phase vocoding-derived components provide thin, soft and variously modified versions of a series of source tones. (1:11) These shadowy constants are punctuated by brief SPIRLZ processed outbursts.
GOSTING
A playful passage from Solo II:
- 64 - A Solo II phrase (s2b) in its original recording. (0:41)
- 65 (- 67) - A collection of three partial representations of the subject phrase (These partial representations were extracted from a different original recording of the same passage.). The three ghostly vestiges have been aligned so as to simulate the effect of the entire original recording. Play: [1] [2] [3] (0:38.5).
- 68 - Here, the "ghostly vestiges" have been carefully aligned so as to match, as closely as possible, the behavior of the complete original (s2b). A visitor with headphones can hear not only the composite alignment of the four components, but also the effect of their spatial independence. Play: Left Channel | Right Channel | Stereo (0:38.5).
STRATA (comb filtering)
Final phrase (s4g) from Solo IV:
WARPNG (alignment differences)
Extended trill passage from Solo IV:
- 70 - Original passage from Solo IV (including s4b1, s4b2, s4b3, s4b4, and s4c) (1:40)
- 71 - A set of four extracts from the previous passage in succession. (0:44.5)
- 72 - The above four extracts (71) are joined into a new continuity. (0:41.5)
- 73 - Now the 41" compo64 - site is extended by phase vocoding. (1:49)
- 74 - An experimental mix of four different stretches so that their various misalignments can be experienced. (1:52)
SPACE
Three aspects of spatialization are demonstrated: A) virtual paths in quadraphonic space, B) a surround-sound immersion involving differing versions of the same materials, C) four independently created channels combine to achieve a complex antiphony.
A) Syrinx SPIRLZ
- 75 - The subject phrase (6"), processed in a 6-cycle SPIRLZ version is then spatialized in a clockwise ellipse, using the space unit generator within the cmusic software environment. This is a stereo mix-down of the result. (1:56.5)
- 76 - A binaurally encoded simulation of the quadraphonic original. (1:56.5)
B) Comb filtering
- 77 - The surround-sound effect of a comb filtered phrase. (1:14.5)
- 78 - A binaurally encoded simulation of the quadraphonic original. (1:14.5)
C) Algorithmically processed fragments (static, positional antiphonies):