Vol. 25 Issue 4 Reviews
Christopher Dobrian: Artful Devices, Music for Piano and Computers

Compact disc, EMF CD 018, 2000; available from Electronic Music Foundation, 116 North Lake Avenue, Albany, New York 12206, USA; telephone (+1) 888-749-9998 or (+1) 518-434-4110; fax (+1) 518-434-0308; electronic mail emf@emf.org; World Wide Web www.cdemusic.org

Reviewed by Daniel Hosken
Northridge, California, USA

Artful Devices contains six pieces by Christopher Dobrian, each of which incorporates a substantial degree of computer performance. The composer states that “the intention is not for the programs to emulate traditional musical behavior, but rather to explore new musical possibilities implicit in certain paradigms of computation or computer decision-making.” Although it’s not clear what “traditional” behavior is avoided (or at least not sought), each piece does realize a different mode of computer decision-making and subsequent or concurrent computer performance. Three of the works utilize algorithmically generated scores which are then performed by a computer-controlled piano or, in one work, a Music-N-style software synthesis language. The other pieces each incorporate a live performer who generates data for a computer program to operate on. The performance paradigms range from the seen but unheard guitar-controller performer to the traditional electronic concerto for MIDI-equipped piano and computer-controlled electronic sounds. There is a sense of play present in most of these pieces and even the cover art playfully reveals the phrase “art vices” within the title.

The first piece on the CD is Entropy, “performed by a Macintosh computer playing a Yamaha Disklavier piano.” In Entropy, “the algorithm takes as its input a description of some beginning and ending characteristics of a musical phrase, and outputs the note information necessary to realize a continuous transformation from the beginning to the ending state.” Although the composer emphasizes computer decision-making in the opening remarks of the liner notes, what makes this piece really work are the decisions made by the composer about the parameters of each phrase and the succession of phrases. Mr. Dobrian engages here in a kind of meta-composition that is becoming a familiar paradigm not only in algorithmic composition but also in the “remix” phenomenon in popular music.

The opening gesture of the piece is one of scattered pointillistic notes in a nervously varying rhythm that builds from single notes to chords. Those chords give way to wide-ranging linear runs in a steady rhythm which then dissolve, creating a cadence point before moving on to the next phrase. This is a very satisfying juxtaposition of materials and the temporal scope of each phrase component feels appropriate. Satisfying successions such as this one are the piece’s strongest attribute. Although it is possible to discern an overall shape of sorts, with specific gestures and phrases returning later on in the piece, Entropy seems to work best when one focuses on the local changes on the level of the two- or three-minute section, letting the work flow by with a general appreciation of the return of materials. This is true of many works whose rhythmic or harmonic grammar is determined probabilistically. This is not a bad thing, but it does prohibit the kind of motivic listening that one usually engages in when experiencing a work which sets out specific motivic or harmonic relations whose manipulation and transformation become the business of the piece.

The second and fourth tracks are two versions of the same work, Unnatural Selection, with the second being approximately half as long as the first. The separation of these two versions by an intervening work is a smart programming move as it forces the linear listener to hear the second version with fresh ears. This despite the fact that the intervening work is anything but a sonic sorbet!

The first version of Unnatural Selection features the composer playing a MIDI guitar providing information to a genetic algorithm which in turn improvises the computer-controlled response. The most interesting part of this performance model is that the direct output of the MIDI guitar is not heard, only the computer’s response to the generated data. Where pieces for loudspeaker performance break the traditional relation between physical gesture and perceived sound by removing the physical gesture of the performer from the stage, this version of Unnatural Selection in concert would present the physical gestures of the performer and potentially only distantly related sonic results. Complicating this is the fact that we are listening to the piece in pure audio form on CD. With more traditional recorded works, we can imagine the physical gestures of the performers from the perceived sound, but here there is nothing in the sound that allows us to mentally generate the performer’s gestures. You can’t play air guitar to Unnatural Selection! While this is not an inherent flaw, it does highlight the cognitive complications of this art form. Although program notes are often derided (“if you have to explain the piece, then…”), they are essential in interactive computer music to give the listener a framework within which to appreciate the behavior of computer and human performers.

Unnatural Selection begins with high-register shimmering sounds that are articulated by diffuse melodic and rhythmic characteristics. In fact, it’s initially quite difficult to distinguish between rhythm and timbre. These sounds develop into a marimba-like timbre that is articulated by an irregularly accented, fast underlying pulse. As the register drifts down to the lower middle register, piano and drum machine timbres are added and the harmonic material becomes more traditional. When the saxophone and bass sounds are added one can, for a moment, imagine the gestures performed on the guitar which might produce such sonic output. A “drum solo” is followed by the addition of bell-like tones and less traditional harmonic material with a continued fast and irregularly accented pulse. The register of the piece again moves upward, and the fast pulse that has been with us since the first few minutes of the piece dissolves into “quarter-note” pulses articulating vibraphone notes in a sea of glassy timbres. The pulses disappear and the piece ends in a virtual standstill on a lovely, if somewhat old-fashioned, shiny timbre that is similar to the opening sonority without the internal rhythm. The standard instrumental timbres present in the piece could be more closely connected to the live performer than is described here, but it is difficult to tell from the CD alone. Despite their inherent allusion to an unrelated instrumental world, these instrumental timbres don’t particularly offend, with the possible exception of the drum machine. At this point in time, it is very difficult to hear canned drum sounds as solid sonic citizens of a piece like this. It might have been less jarring to hear percussive sounds that aren’t quite so imitative.

The piece has a clear overall high-low-high registral shape that unfolds at a leisurely pace despite the rapid surface activity. As with Entropy, the various sections of the piece don’t provide much local motivic material, just general statistical behavior. And again this works quite well if one is prepared to listen in that way.

The second version of Unnatural Selection gives a more readily cognizable situation where an improvisation by Daniel Koppelman on a MIDI piano provides the data for the genetic algorithm. This version has a similar overall shape and succession of timbres but is compressed into about half the time. The pianist’s free improvisation frequently has a jazzy flavor giving this version of the work a different sort of feel and rhythmic energy. The performer also provides the kind of localized motivic material that is avoided in previous pieces on this CD. This is a nice contrast to the unheard MIDI guitar version, reinforcing the value of including two versions of the same piece. The saxophone and drum machine timbres play a very limited role here which suits this combination well. The long registral arc of the first version is less clear here but that is more than made up for by the additional musical interest of the piano improvisation. Mr. Koppelman’s improvisation is well structured in terms of texture and rhythm and interacts quite well with the program.

The piece that intervenes between the two versions of Unnatural Selection is Line/Phase Minutiae performed by computer-controlled Disklavier. This eleven minute piece is an “assemblage” of one-minute computer-composed sections. These consist either of repeated notes in harmonically related tempi each of which fades in and out or of “florid melodic figures” that were created “in real-time according to destination points indicated by the human composer.” The limitation to the two types of textures seems to be an overly limiting factor in this piece. For a few minutes, the succession of one-minute fragments is attractive, but the pacing of the juxtapositions becomes too predictable and the alternation of textures loses its interest. There are, of course, variations in register, tempo, and so forth, but that doesn’t stave off the unfortunate static quality of the textures. The two versions of Unnatural Selection feel very fluid and fresh by comparison.

Degueudoudeloupe is a six-minute piece composed using algorithms written in C and performed by cmusic, a Music-N-style sound synthesis program (running on a VAX computer, no less!). The persistent, irregularly-accented pulse that appeared in the previous work is present here as well. The timbres are largely percussive and inharmonic and are less lively and reverberant than the other synthesized timbres on the CD, which sound as if they were produced on commercial synthesizers. This gives the piece its own timbral space, but it feels a little too dry by comparison. Degueudoudeloupe sounds like a study and lacks some of the lively drive of the other works. Nevertheless, there is a discernable shape and the work doesn’t overstay its welcome.

The final work, There’s Just One Thing You Need To Know, features Daniel Koppelman, again on Disklavier along with a Korg Wavestation A/D controlled by Max. The composer describes the “conceptual and music theme” of the piece as “reflection.” The piano part requires symmetrical hand movement and the music played by the computer is “symmetrical around a given pitch axis.” The computer’s response to the pianist is a mix of the techniques used in the previous pieces ranging from accompanist to improviser.

The work opens with glassy-textured, filter-swept synthesized sounds responding laconically to the pianist’s few notes. This relaxed sonic bath quickly turns to a more tumultuous texture as the pianist and synthesizer begin churning out notes. As with the previous tracks, the music here works by juxtaposing phrases of opposing textures and musical activity. The chunks often seem a little bit too short and/or too regularly paced, making it difficult for the listener to identify an overall “sweep” to the music that the local gestures and timbres would seem to favor. Harmonically, the piece often finds itself circulating through various materials without a clear sense of how these are operating structurally. The piano writing is generally quite satisfying and the restriction of the music to symmetrical gestures doesn’t seem to inhibit the pulsing, syncopated gestures that characterize most of the music on this CD. The timbres here are usually spacey “electronic” timbres that serve the work just fine but are otherwise unexceptional. In this piece I enjoy many of the moments, but as a whole it doesn’t leave a clear imprint.

Artful Devices is a disc of mature works that combine to form a clear picture of the composer’s musical concerns in this genre. The performer–computer and composer–computer interactions are clearly of primary importance for Mr. Dobrian, and these works present a wide variety of those interactions with a good deal of musical success. Unlike much computer music, the composer doesn’t seem as interested in timbral innovation—he appears content to skillfully manipulate “stock” sounds to form a coherent timbral shape for each work while concentrating on the musical structures generated by the human–computer interaction.