Conference napalm::commusic_v1

RE < Note 2171.0 by BAVIKI::GOOD "Michael Good" >

	I'm not sure I understand your objectives, but regarding the choice of
synth modules to use, the D5 is probably a good start.  It can produce many 
different simultaneous sounds and has built in digital reverb.  I was all set to
recommend an MT32 before I read of the D5, which is almost the same unit, with
the exception of the MT32 unit not having a keyboard.

	Dan

    I would imagine that an analog rackmount synth, rather than various 
    digital technologies, would be more controllable - and it would have to
    understand (via MIDI SYSEX) 
    
>      - Multi-sound and multi-timbre
    
    Various Roland analog synths meet this criteria - also the Oberheim
    Matrix-1000 (unless that was 6-voice monotimbral, I forget)
    
>      - Realtime continuous control over loudness, pitch, timbre, etc.
    
    Micahel, remember that via MIDI you can generate requests for discrete
    pitches, usually based on an A440 12-note scale, with a limited amount 
    of pitch control, using the pitch-bend controller (CC#1).  The 'timbre'
    selection would be a standard MIDI patch change request, and the
    'loudness' would be a function of MIDI volume (CC#7) and transmitted
    velocity value (0-127).
    
    Roland MKS-7 ? len ? Edd ?  
    
>      - Processing (e.g. filtering, reverberation) as well as synthesis
    
    Filter cutoff has a MIDI controller number, I beleive.  For
    reverberation, there are only a few units that have integral effects,
    and those are of the 'audio workstation' variety.  Okay, there's a
    couple...
    
>      - 5 ms or less response time for audio cues (a function of the
>        computer/synthesizer system, I expect)
    
    I dunno if you're gonna get this response out of a MIDI unit, because
    of the limited MIDI serial bandwidth and the delay (if any) from the
    synth's opto-isolator MIDI port... any o' you hardware engineers want
    to continue this ?
    
    BTW, at a customer site I was given a demo of a SUN Sparcstation - it
    allows voice sample files to be played at the desktop thru a small DAC
    and speaker, both builtin to the unit.  
    
    karl
    

    also, according to Steward Brand's book, the MIT Media Lab has been
    doing research projects on this area for years..
    
    karl

This is great - lots of help already!  Thanks very much.

Re .1 - If the D5 is basically the same as the MT32, plus keyboard,
that's good to know, since the MT32 was one of the units the tutorial
recommended (the other synthesizer units being the Roland D-110 and
the Yamaha TX-802 and FB01).  The keys are probably useful to us
for some early demos, as an easy input device.

Re .2 - I'm very new to the MIDI game, most of my knowledge coming
from the limited description in the tutorial.  So your pointers relating
MIDI features to my list of requirements are helpful.  

I'm not sure I understand the point in the first paragraph, though.  
What are examples of Roland analog synthesizers?  How would they be 
more controllable than digital synthesizers?  Does this still hold true 
if we want to add on sampling capability to our synthesis at a later date?

Re .3 - I'm a graduate of the MIT Experimental Music Studio (one of the
groups that became the MIT Media Lab), so I can vouch for that group
doing research in computer music for over 10 years.  I'm acquainted
with Vercoe's and Machover's recent work, but while it's related, it
isn't really covering what we're doing (same story in the visual areas;
the Media Lab is doing interesting work, but it's less tightly related
to ours than it might seem at first).  Real-time control, of course, 
was not what we were working on 10 years ago.

    If you want smooth continuous control of timbre, you might want to go
    with an FM synthesizer instead.  It depends what kind of noises you
    want to create.  Just musical ones?  Sampled real clunks and voice?
    Continuous washes of sound with tonal coloration carrying meaning?
    (New Age music as an output metaphor?  ye gods.  I want to see the
    "UI style guidelines" book for this.  Only 18th century counterpoint
    allowed?  DECmusic - only corporate-approved socialist realism?)
    
    Isn't your group the one that has a Next workstation?  Those things
    can play synthesized sounds directly, as can any old Macintosh (to
    varying degrees of quality, depending on which model you've got.)  The
    Next machine can do CD-quality output, but you have to program all the
    effects yourself.  Unless somebody has already done that for you.
    
    If you want FM, there are some Yamaha units with some built-in reverb
    effects.  You probably do not need something with a keyboard.
    
    At full bore, a MIDI signal can convey over 1000 events per second.
    Your major bottleneck may turn out to be the VMS QIO routines.

    Most personal computers that have audio cues have a DSP capability
    (NEXT) or at least DACs on board (Amiga), or some other specialized audio
    chips (ensonique on atari?).  MIDI seems an awkward way to add sound to
    computers for general cues.
    Tom

    I see the MIDI connection as a shortcut - avoiding the necessity of
    developing a DSP board for - VAX ? RISC ? what bus ? SCSI ? async or
    sync SCSI ?
    
    The reason I mentioned analog synths rather than digital (FM, etc) is
    that I have a naive idea that filters, envelopes, VCA, etc. would be
    easier to control via MIDI than having to really, really understand
    Yamaha FM, for example, and have to xmit (using SYSEX?) these numerical
    values to the digital synth.  Feel Free to cut me off at the knees,
    everyone.
    
    karl

Not really knowing what I am talking about, I'd agree a bit with Karl --
an analog synth would be an expressive way to go.  With FM I'd go
TX81Z and not FB01 and with Roland D-110, not MT32 (and the D110 is
readibly available on sale for $600 if cost is any matter).  Anyway,
I want to see the demos when your done!  When do you have the Babbage
scheduled??? :-)

Chad

gee,

somebody who gets to read commusic for his job :-)

Chad

The TX81Z would be a good choice except that FM is tough to deal with.  Not
just because it is FM, but also because Yamahahahaha does not implement FM
exactly per theory.  Seems to me you have to go with a Fairlight, is it, if
you want "pure" FM?  Anyway, I suspect that the best type of unit to go with
would be a sampler.  This way, if you have any sounds that you want to try and
can get a decent recording of them, you can load them in and try them.  With
an LA, FM or analog synth, you are going to have to fiddle around and be
satisfied with something that is "close", assuming you have become familiar
enough with the beast.  I would probably go with a 16-bit, stereo sampler,
like an FZ-10, FZ-10M or TX16W.  They can be had for, what, about $1200 or 
so?  Seems to me that would be a better choice than limiting yourself to 
one or another form of synthesis.

Steve

The TX16W is stereo but 12 bit (sounds good though).

Chad

Thanks again for the help!  Here are some answers to some of the questions
that have come up.

Re .5:  Yes, we have a NeXT (and several Macs), but the goal here is
        to build prototypes based around DEC workstations (VMS or UNIX,
        VAX or RISC).

Re .8:  I don't think we'll have a Babbage presentation at Spit Brook
        until sometime after New Year's.

Re .10:  Yes, I think we'll want to use both sampling and synthesis.
         Thanks for the recommendations in both areas. 

To the folks who reply to this note - next Monday I would like to extract
this note and post it in the Presence conference.  Please let me know
if you have any objections to your reply being posted there.

	As a side note, the D5 doesn't have built in reverb, where the
	D10 (similar set up) and the D110 does. The MT-32 does have built in
	reverb. The built in reverb makes a large difference on the drum 
	sounds.

								Jens

BTW, can't you get something that will allow the NeXT to be played via MIDI?
Seems that would be a good option.  You could get RS-232 to MIDI hookups if
needed for both the VAX and the NeXT, or something like that I think.  Once
you get a sound you like, you can go get the synth that does that sound.
Seems to me the demo from a year ago included having the NeXT run as a CD-
quality sampler.  

Sorry about the -bit error on the TX.  Easy mistake to make when you got no
gear... :)

Steve

Re .14:

If it was just our group in Spit Brook putting together the demos,
then that might be an option.  But we have a more complicated set
of goals.

Chauncey Wilson's been demoing the NeXT's sound capabilities as part 
of our group's competitive analysis presentations, but even though 
we already have the machine here, I don't want to work with it as an 
integral part of this project.  For one thing, the group in Albuquerque
doesn't have one.  The main reason, though, is that about a year
from now we want to put together a parts list and a software kit
so that people throughout Digital can build these prototypes around
their workstations.  A NeXT isn't cost-effective in such a kit.
But it's a dynamite machine in other ways.

However, we can net the NeXT and our Macs to our DEC machines, so
we could use software packages written for either machine to develop
(as opposed to run) our demos.

    You may find that it is more cost effective for someone to go out and
    just buy an off-the-shelf Mac than to buy other specialized hardware and
    try to interface it to a VAX workstation.  MIDI-controlled samplers are
    not cheap, and a VAXstation's interface capabilities are quite limited.

    I'm not completely sure I understand what you're looking for (or even
    looking to do).

    If you're looking for instrument reproductions and "musical sounds"
    (eg, for computer-based audio/visual presentations), then any synth
    with "hip" sounds will do.  If, on the other hand, you're looking for
    sound *effects*, a simple analog synth is probably easier to program. 

    (Like I said - I don't understand your application completely.)

    Anyway, as long as you're considering SGUs (sound generation units),
    then you may want to take a look at the new Roland boxes, which are
    designed for computer usage (as opposed to front-panel tweaking).
    They're called: 

	CM-64	64 voices, L/A synthesis and PCM samples
	CM-32P	32 voices, PCM samples
	CM-32L	32 voices, L/A synthesis (basically a re-boxed MT32)

    Roland also has a PC plug-in L/A sound card called LAPC-1.  Don't know
    if this will help or not . . . 

-b

    Okay, how about this ...  Pick a sampler, any sampler taht supports
    MIDI Sample Dump Standard and do your development on it.  Then, later 
    on folks will grab your software and the recommended hardware that adds 
    a MIDI port.  They will also buy the sampler of their choice, dump your 
    samples into their sampler and be off and running.  That way, you are 
    almost guaranteed that your work will not become obsolete by the time you
    come out with it.  
    
    For example, suppose Roland comes out with a souped up, but incompatible, 
    version of LA synthesis.  You did development on a D-110 which Roland 
    discontinues in a year.  You are SOL.  But, if you use, say, a TX16W 
    for your development and use Sample Dump Standard, and they discontinue 
    the box, you should still be able to buy another box a year from now 
    that supports the standard.  
    
    Steve

Re .18:  Interesting point about upward compatibility differences between
samplers and synthesizers; something I had not thought of before.

Although lots of alternatives were suggested, the consensus here seemed
to be that the D-5 was an acceptable choice of synthesizer.  That's what
we got, so we have compatibility with the Albuquerque group.  We'll be
getting a sampler later.

A couple of groups at universities are working on (or have already developed)
software which works with MIDI devices and DEC/VAXstation 3100 computers.
I believe the group at University of Lowell is using an RS232-to-MIDI
converter made by Hinton Instruments in England.  The MIT Media Lab may
also be working on this.  DEC funds both the Media Lab and the Lowell
graphics lab, so we're in a good position to use and test whatever they
develop.

Re .17:  A big part of the project is to work together with customers to
see what applications can make use of which parts of multi-sensory I/O
technology.  Two possible applications: 

1) a musical complement to scientific visualization for multi-variate data, 
where multiple dimensions are output via audio as well as video.  This is what
the U. Lowell graphics lab is working on.

2) an audio component to user interfaces which uses everyday sounds as 
metaphors, much as workstations try to use everyday pictures as metaphors - 
earcons as opposed to icons.  Apple has done some work here with their
Sonic Finder prototype.

Thanks again for the help!  I should be hearing the D-5/MicroVax II demo
this afternoon.