source file: mills2.txt Date: Sat, 6 Jan 1996 11:33:49 -0800 From: "John H. Chalmers" From: mclaren Subject: CLM, Linux, and microtonality --- In graciously responding to some of my gripes about Csound, William Schottstaedt mentioned that his CLM emvironment runs under Linux. This was news to me. Since CLM is by all accounts one of the most deluxe composition/ synthesis environments, it's welcome news indeed. This forum has, since its inception, served as hang-out for 2 kinds of subscribers: ordinary schmucks with IBM or Mac desktop systems and mostly MIDI synths and software, and the few, the rich, the tenured--who generally use some flavor of UNIX (NeXTstep, typically) to run software-based synthesis apps like CMix, CLM, or Csound. Relatively few of the ordinary yutzes (like, say, moi) compose much with software-based synthesis languages. In part, this is because it's so much faster & more efficient to use MIDI. In part it's because the academic freeware is so damn hard to use and almost completely undocumented, but let's not beat *that* dead horse. This leads to a huge echoing chasm twixt the academic microtonalists and their dirt-poor scumsucker "just folks" counterparts. The two groups literally speak different musical languages. MIDI is a dumb, slow protocol that basically tells dedicated hardware when to goose-step. 99% of MIDI's messages revolve around modulation--vibrato, filter cutoff, tremolo settings, reverb depth, envelope bias, etc. There is no MIDI continuous controller dedicated, for example, to making a timbre more inharmonic or adjusting the Nyquist rate of the synth. MIDI commands reflect this concern with note start times. There's a note-on, but no message that tells the synth how long the note will be--a clear indication of MIDI's design as a real-time performance protocol. 90% of the parameters in all MIDI synth patches control real-time modulation--response to aftertouch, vibrato depth as a function of wheel position, attack rate as a function of key velocity, etc. MIDI is a coarse-grained protocol. If you send more than 20 or 30 note-ons, you'll get arpeggios instead of chords. In contrast, the software synthesis languages like CLM, Csound and CMix are relatively smart and very finely granular. While MIDI doesn't know or care how a synth responds to a note-on message, software instruments can change their behaviour depending on the note's pitch, its duration, the number and type of other notes playing at the same time, the location in space or in the overall composition at which the note appears, etc. Software synthesis instruments are typically specified at the level of the individual overtone and involve no real-time modulation parameters. Computer composers often write programs to explicitly generate dozens or even hundreds of parameters for each software-synthesized note outside of real time, so responsiveness to real-time modulation parameters is a non-issue when composing in Csound or CLM or CMix. The parameters are "built in" for each individual note and can be exquisitely fine-tuned. This has important consequences for microtonal composition. Software-synthesized xenharmonies tend to be very finely wrought, with layer upon layer of acoustic complexity. Software-synthesis microtonality appears to be centered as much around timbre as around notes. Even in the works of those composers notable for their contrapuntal skill (Paul Lansky, William Schottstaedt, Bill Alves, Jonathan Harvey) timbre remains uppermost as a factor in the "non-12" sound of the music. By contrast, MIDI microtonality centers almost entirely on harmony and melody. Timbre is of minimal concern, because commercial synthesizers offer nothing in the way of detailed fine-grained control over that parameter. Retuning individual overtones remains outside of the purview of the MIDI composer, *especially* during the course of the composition. (William Sethares has written some custom FORTRAN routines to read LEMUR analysis files and warp FFT'd sounds into a given tuning, but this is a rare exception, takes gobs of time, and demands a sampler with enormous amounts of RAM. Thus Bill Sethares remains the lone exception in this regard.) The very model of the MIDI microtonal composer is Warren Burt, whose compositions are essentially contrapuntal and harmonic. Timbre is a non- issue. This gives an interesting 18th-century "gebrauchmusik" sound to Warren's compositions, even though they use rhythms, melodies and harmonies no 18th-century composer would ever contemplate, while it gives the compositions of someone like Paul Lansky an oddly modernistic sound--despite the fact that Lansky often uses conventional 12-tone equal temperament (as in the "idle chatter" series of compositions, all in g minor, whitebread 12/oct). To date there's been little contact twixt the two camps of microtonalists. Thus, many MIDI xenharmonists confidently make statements about timbre, consonance and dissonance which are simply untrue when software synthesis (allowing individual partials to be retuned) is involved. In like manner, many academic microtonalists lose sight of the musical forest by exploring mathematical partition-function and pitch-class set/chaotic note generator minutia, rather than asking the larger questions: What kind of intervals does this tuning have? How does it "sound"? How can harmony and melody be used in this tuning in ways which differ productively from harmony and melody in 12/oct? Thus, the investigations of John Clough, Gerald Balzano and Carlton Gamer appear to have had much more impact on MIDI microtonalists (for example) than on software synthesis composers. Meanwhile, the ideas of folks like William Sethares and James Dashow appear to have percolated more thoroughly into the software synthesis camp than the MIDI contingent. This has led to further confusion and miscommunication. Academics often write some of the most insightful discussions about the internal structure of non-12 tunings, while non-academics often write some of the most useful monographs on the "sound" of non-12 tunings and the interaction of tuning with timbre. NeXTStep might have bridged the gap twixt these two microtonal factions--but alas! That operating system is now dead and buried. It's been overpriced FAR out of reach both of academics *and* the rest of us, and is now essentially defunct except on antique legacy machines like the NeXT cube. (A machine roughly 25 times slower than the P6 @ 200 Mhz.) Recently, however, Linux appeared...and this operating system might finally offer a bridge twixt the two worlds of microtonal composition. Linux is extremely stable, according to my UNIX- guru friend. Under X Windows, it's reportedly easy to use. The big stumbling block right now appears to be setting up Linux to run under X Windows with your particular monitor... The process is more complex than superstring theory. (Do YOU know the "dot rate" of YOUR monitor? Not me!!) Only within the last 5 years have desktop IBM machines grown fast enough to fully shoulder the burden of software synthesis. But now that it's happened, IBM PC prices have dropped so far so fast that it's hard to imagine anyone in or out of academia will be using NeXT cubes or other legacy antique machines for very much longer. The InfoMagic 4-CD set of Linux with complete X Window support and all utilities now runs a whopping $20 down at your local software store. It's very hard for me to believe that NeXTStep-486, at a cost of 5 thousand dollars (yes, $5000.00), will survive the competition with the $20.00 Linux operating system. Linux can read DOS disks and apparently offers the user the option of setting aside one partition of the hard disk for DOS files. This clearly would go quite a ways in bridging the gap between software and MIDI synthesis. Ideally, as a microtonal composer, I want total control over EVERY asect of the composition-- and reasonably easy, fast, efficient control. Combining Linux-based spectral modification programs with CMIX, a sampler, and MIDI sequencers might accomplish this. My ideal system would let me tear apart an acoustic sound, retune the individual partials, then generate real-time scores with the ease and simplicity of a MIDI sequencer. Obviously this goal lies some years in the future, but the Linux and DOS/Windows combo seems at least a step in that direction. Right now this kind of integration is essentially impossible. Even the NeXT cube didn't offer MIDI sequencing with audio synchronization, nor analysis with real-time resynthesis. But now, with Linux, there might finally be a bridge between these two styles of microtonal composition. Especially if & when someone produces a microtonal synthesis program like James McCartney's SuperCollider that runs in real time on an *affordable* computer (the Power Mac is not currentlfordable and probably never will be--everyone with $4000 to spend on a Power Mac raise your hand, please). Such a program would bode well for interactive real-time (those 90s buzz-words!) acoustic-and- digital microtonality. --mclaren Received: from eartha.mills.edu [144.91.3.20] by vbv40.ezh.nl with SMTP-OpenVMS via TCP/IP; Sun, 7 Jan 1996 00:16 +0100 Received: from by eartha.mills.edu via SMTP (940816.SGI.8.6.9/930416.SGI) for id PAA00718; Sat, 6 Jan 1996 15:16:46 -0800 Date: Sat, 6 Jan 1996 15:16:46 -0800 Message-Id: Errors-To: madole@ella.mills.edu Reply-To: tuning@eartha.mills.edu Originator: tuning@eartha.mills.edu Sender: tuning@eartha.mills.edu