Conference napalm::commusic_v1

43.0. "Colgate Studio, DMX-1000, ..." by KIRK::BOYER () Wed Nov 28 1984 21:22

I was very pleasantly surprised to find a notes file for computer music
here on the E-net.  Having just transferred from the wilds of upstate New York
to the mill (I used to work for software services in Ithaca), I am not used to
such luxuries.  What follows is a history of a computer music studio I have
been intimately involved with, and my specific involvement in computer music;
I'm hoping some of you might find areas of communality or interest: 

It all started at Colgate University (again in upstate) where, from approx.
1974 -> present a very state-of-the-art computer music studio has existed, run
by Dexter Morrill who is a professor of music there. During my attendance, the
CPU in use was the reliable KA10 (PDP-10 system) running all the latest
Stanford software (Leland Smith's SCORE, MUSIC-10, etc...).  Dexter had/has
numerous ties to Stanford (close friends with Chowning, Smith, Moorer, and most
of the others who, at one time, hung out there), thus the system at Colgate
closely resembled what was happening at Stanford.  Many pieces were composed at
Colgate by students and, more often, by visiting composers.   The system was
very non real-time in the sense that the MUSIC-10 compiler chugged along
computing the samples for each note of the score (during normal PDP-10
timesharing) and outputting the numbers to a disk file.  A 10 second section of
music could take 1 hour of compiler run time to generate the samples.  Tough
luck if (as was almost always the case) you didn't like the result. Then,
another program (PLAY) was run to read the file in real-time and dump it to a
12-bit D/A converter (custom built).  This system was admittedly cumbersome to
use but was very general; allowing very complex instrument definitions, large
arrays for reverberation and complex envelopes, and lengthy scores.   Dexter
composed perhaps a dozen fairly well-known pieces using this stuff, his most
famous being 'Studies for Trumpet and Computer'; one movement is called
'Rotations' and consists of a quad speaker set-up playing computer generated
sounds moving spatially while a (human) trumpet player stands in the center and
also physically turns while playing along.  The result is an incredible
auditory experience.  If anyone wants a copy of this piece I could probably be
bribed into making a cassette... 

During my college years at Colgate and the following years, I managed to
complete about six compositions, my most successful being a piece for 'live'
cello and computer generated tape (Illusions I).  A piece written for flute and
computer (Illusions II) was performed at the International Computer Music
Conference at Eastman in Rochester last year.  Somewhere amidst writing these
pieces I graduated with a degree in music (very practical) and landed a job
with Digital.



Alas, Colgate finally went the way of many other universities and, in 1982,
converted from PDP-10 to VAX.  Instead of trying to compete with everyone
else in the school for machine time, Dexter decided to go to smaller systems.
He purchased a PDP-11 with a 40mb Winchester running RT-11 and a special
'music processor' known as the DMX-1000.   The DMX-1000 was designed by an
MIT graduate named Dean Walraff, and he has formed a company in Boston called
'Digital Music Systems' (some of you might have seen the adds in Computer
Music Journal).  The system looks like this:


	+-------------+
	|             |           
	|  PDP-11/23  +-----+     +-----------+
	|  28 Kw      |DMX  |     | DMX-1000  |
	|  2 x RX02   |int. |=====|       +---+
	|  1 40mb Winc|     |=====|       |A/D|
	|   (non-DEC) |100hz|=====|       +---+
	|  RT11 sw    |     |  ^  |       |D/A|
	|  Terminal   +-----+  |  +-------+---+
	+-------------+    ^   |   
                           |   |
                           parallel 
			   interface

The DMX-1000 has memory for 256 microcode instructions which are executed up to
approx. 65K times/second.  The microcode is quite dense (many operands) allowing
many software instruments to be packed into the DMX for 'real-time' playing. A
compiler is provided to run under RT11 on the 11/23 which translates
MUSIC-11-like orchestra definitions into DMX-1000 microcode. Then, the
microcode is loaded into the DMX via the parallel interface.  The DMX also
contains memory for functions (envelopes or waveforms) which are also down-line
loaded.  Then, a score is created (on the 11/23), sorted into a sequential note
list file, and stored on a disk.  A program provided by DMS called M1000 then
reads this file and uses the data to update the DMX via the parallel interface
in real-time. 

One nice feature of the DMS compiler is that all the orchestra definition
units (software oscillators, filters, etc) are defined via MACRO-11 macros;
and thus the instrument designer can use any MACRO instruction within the
instrument.

Omitted from the above diagram is another PDP-11 <==> DMX interface which
allows high speed DMA transfers to/from PDP-11 memory.   Using this interface
and the A/D unit built in to the DMX, 'live' sounds can be sampled by the DMX,
passed to the PDP-11, and stored on the Winchester.  We almost have the
software for this debugged (it works but it's a little noisy!).  We're hoping
to add a digital mixer and some analysis software in the near future.  

Since the DMX uses a 16-bit D/A (and A/D) at 65K, the sound quality is
excellent.  If anyone is interested in more information about this system,
let me know.  I've got most of the DMS documentation and some useful instrument
definitions. By the way, the DMX-1000 sells for about $10,000 the last I heard.

(Using this system, Dexter Morrill has just finished a composition for 
saxophonist Stan Getz which was recently premiered at Stanford and has 
already had several performances in Europe.)

One nice recent addition to the Colgate studio is a SONY digital tape
recorder (uses video cassette deck with an extra box of electronics).  We
like to daydream about interfacing this unit to the DMX-1000 (the problem
is 2 incompatible sampling rates).  In any case, this is a real nice machine
and cheap (less than $1000. I believe),  considering digital tape decks used
to cost $100,000 or more.

This is the current status of the Colgate Studio.  The last time I talked to
Dexter, he indicated his DX7 would soon arrive.  As to how we'll interface
to it we're a little confused at this point.  To confuse the issue, we have
easy access to a VAX750 running UNIX and we're wondering about the
possibilities of mixing/editing/analysis of sound files (from the DMX or DX7?)
on this machine. 


Areas I'm particularly interested in are:

	Spatial sound movement
	Interfacing the DX series to PDP-11 or VAX (does it make sense?)
	Digital recording/reprocessing of 'natural' sounds
	Digital mixer programs
	Starting a computer-music division at DEC, or obtaining
		(free of charge, of course) DEC hardware for 
		computer-music systems (I like to dream)

If anyone cares to communicate I'd be greatly obliged.

Rob Boyer    @CIM Engineering    DTN 223-5005
(on the E-net at KIRK::BOYER)

Interfacing a Yamaha DX7 to a PDP-11 should be straightforward.  You can 
buy interfaces to the Apple II and the IBM PC, which differ only in the 
interface; a little hacking should produce a Q-bus interface.  Look into 
the Roland MPU-401--that's what I use.

Although I used to be "one of the gang" at Stanford I missed your friend, 
since I left in late 1968.  I wrote part of the software that drove 
Stanford's music DACs; did Colgate use that same device driver?  It had 
provisions for sharing the channel with a Type 30 display ripped off a 
PDP-1.

I understand that Andy Moorer is now at Industrial Light and Magic.
    John Sauter

Thanks, John, for the info on interfacing the DX7.

I'm not sure what driver we used at Colgate to drive the DAC, but it was
most probably yours since there didn't exist a lot of high-powered
programmers at the time.  The DAC itself I believe was designed by
a guy called 'Zingheim' or something like that who they flew in from
Stanford to do the job. Being a music major, I didn't get into the
internals of the DAC driver or the DAC itself -- I never heard mention
of the Type 30 display....

What is Industrial Light and Magic?  Have you ever heard any of Moorer's
pieces done at IRCAM?  I've got a few somewhere and they're definately
worth listening to.

Rob

Industrial Light and Magic is the special effects crew that is responsible
for many of the "space" movies' effects, such as (I think) Star Wars and
Star Trek.

	- Rick

P.S  While you're working on it, how about a Rainbow MIDI box?

I don't recognize the name Zingheim, he may have been after my time.  I've
never been to IRCAM, and I've never heard any of Moorer's work.  Yes, ILM
is the group that did Star Wars and the Star Trek movies.  Maybe I've heard
Andy Moorer's work when listening to those movies?! 

The Type 30 display was DEC's first computer display.  It had a 1024 by 
1024 resolution screen, and was available in the early 60s.  We used the 
music DACs to drive it, so there was a feature in the driver that would
blank the display while music was being displayed, then restore it.  
Whoever was using the display would listen to the music while his work was 
being interrupted.  For long pieces (more than a few seconds) the music 
program would take over the whole computer so it could read the next buffer 
from disk in time to keep the DACs running.  At those times everyone would 
listen to the music.
    John Sauter