Conference napalm::commusic_v1

    Good stuff Tom, you beat me to it.  Now I don't have to do any
    homework.  ;^)
    
    I doubt it's percussion that's eating up MIDI channels in pop mixes.
    It's more likely the "more the merrier" syndrome.  With 48 tracks
    available, there's a temptation to use them all and then work the
    production issues at mixdown time.  Also, I suspect some folks are
    trying to avoid saturation of MIDI channel capacity by using multiple
    busses, whether or not the problem is real.
    
    Why "old" orchestral scores?  Is all that newfangled stuff timbrally
    minimalist?
    
    len.                        
    
    

    >< Note 1420.1 by DRUMS::FEHSKENS >
>    
>    Why "old" orchestral scores?  Is all that newfangled stuff timbrally
>    minimalist?
>    
>    len.                        

    There are no important new works for orchestra, and havn't been
    for probably 10 years.  John Cage's Cheap Imitation is a wonderful
    work, but has never been performed properly.  It irrevelant to MIDI.
    It depends heavily on the drastic and unpredictable changes that
    occur when the number of instruments on a note is random and the
    choice of instruments on a note is random.  Acoustic instruments
    have such varying volumes (ca. 120 dB range) that MIDI synths couldn't
    do it.
    Anyway, Glass is writing for more standard ensembles now, but he
    is also writing less exciting work, so it doesn't matter.  He'll
    be
    remembered for the early works for a couple synths, sax, vocals,
    etc.
    
    Traditional orchestration has gone the way of the Do Do bird.
    By ignoring new composers, orchestras condemned themselves to
    obsolescence, and helped spur interest in MIDI, which will
    progressively replace the live musicians.
    
    You can safely ignore the orchestra works that receive the Pulitzer;
    no important music has ever won a Pulitzer.
    TOm

And the not-so-important new works are not really using the orchestra in ways
more demanding than the 19th century composers (and early 20th) did.

Leonard Bernstein?  John Williams?  Could be Saint-Saens as far as what
they demand of an orchestra.

So if MIDI can handle the stuff Tom has mentioned, it can probably handle
anything.  (Well, there is that Beethoven piece for three orchestras.
"Wellington's Victory"?  I don't know how big each of the orchestras is,
though.)

Where is the cost/benefit break for synthesizing a string choir?
Should we have one heavily chorused string sound for all registers?
Should we have a sound for each instrument?
We cheated a lot on the above examples.  A string choir should use
17 MIDI channels.

Violin 	E string
	A string
	D string
	G string
Viola	A string
	D string
	G string
	C string
Cello	A string
	D string
	G string
	C string
basses	G string
	D string
	A string
	E string
	optional re-tunable fifth string exceeds 16 channels.

Should we synthesize a group of violins on the E string, or a solo,
and then chorus with a quality choruser?

Are people going to forget what it what it sounds like when a whole section
changes strings at the same time?

Tom

    Well, this is an interesting premise.. how do/would orchestral works
    translate to the MIDI experience ? It's true that a string section
    changing notes is far richer than the best sampled or synthesized
    approximation. It's also true that I don't have 18 string-playing
    friends. And if I did they couldn't fit into my studio.
    
    The massive synthorchestra 'realizations' by Tomita, for example,
    were done on pre-MIDI analog gear with LOTS of overdubbing. Is MIDI
    the proper medium to render complex 'orchestral' music ? Depends.
    One method to minimize MIDI channel usage has not been mentioned.
    While the MIDI spec limits one bus to 16 channels, there is no maximum
    number of notes each channel may play (okay, 128).. rather, this
    is a HARDWARE limitation, i.e. the SGU's polyphony limits. One could
    of course assign one OR MORE SGU's to respond to the same MIDI channel.
    I have several setups for my Emax sampler that have multiple instruments
    across the keyboard.. valid as long as the instruments' ranges do
    not overlap. Therefore a fairly full woodwind section could be done
    completely within ONE MIDI channel. 

    Good topic, Tom     

    I don't see the 16 channel limitation as preventing anyone from doing
    orchestral music.
    
    300 channels would be nice but its simply not necessary.
    
    16 channels limits you to no less than 16 discrete parts on any one
    network.  You are allowed to have more than one network, and using
    splits and mappings you can even put more than one part on one channel.
    Orchestra music is particularly amenable to this.
    
    With devices like the MX-8 going for so little, it seems like a
    reasonable (if not "ideal") way to do things.
    
    I personally would rant and rave if they came up with an incompatable
    MIDI-follow on only to address the 16 channel limit.  I don't see
    sufficient justification for it on that basis alone.  There are plenty
    of reasonable workarounds and if there was significant demand for it,
    I would think that we'd have seen sequencers that handled multiple
    OUT ports (perhaps they exist).
    
    So, what I'm saying is that the 16 channel limit does NOT prevent
    you from doing orchestral stuff.  It is a minor annoyance at best.
    In fact, you will find plenty of MORE annoying problems to solve.
    
    	db

For what it's worth ... 

    C-Lab's Creator sequencer for the ST DOES handle multiple outs, and can
    address up to 64 channels.  (Also, the MC500 can address 32, as can
    Performer [I think].)

    Assuming that you have a unit that can have multiple "split points"
    (like Karl's EMax), and that we have 8 "splits" per board, you are now
    up to being able to address:

	64 inst/chan * 8 inst/box = 512 discrete instruments

    By placing a few MX-8s in the chain, it is possible to address an
    almost limitless number of timbres.  So while MIDI doesn't allow this
    kind of mapping directly, it is still possible.

    BTW - I can potentially address 40 discrete instruments with my current
    5 box setup, not including my HR16.  Not that I ever would, but it's
    possible.

-b

    Only the MC500 Mark II/Turbo MRC-500 combo can handle 32 channels;
    The MC500/MRC-500 combo has two outputs, but they split the usual
    16 channels.
    
    len.
    

    In Elliot Carter's Piano Concerto, going into the second movement,
    there is this big string cluster.  I can't remember the library's
    score perfectly, but it probably about 50-60 notes going at once.
    About Varese's Ionisation, after I studied it last night, I couldn't
    find an instantenous sound with more than 16 sounds or too many
    notes.  The piece on the whole might be possible on a good system,
    but difficult to match sounds (it's totally percussion, all kinds
    of percussion instruments, including sirens, a piano, tam-tams of
    different sizes, etc.).
    
    The Ives concord sonata for piano also has some black-key clusters
    held down with a board, but that's probably only about 15 notes+
    the other lines, around 20-25 pitches.
                                         Tom

    Tom,
    
    Number of simutaneous notes isn't by itself a problem for MIDI.
    Each note is (approx) two MIDI events no matter how long its held.
    
    	db

    re: .3
    
 > anything.  (Well, there is that Beethoven piece for three orchestras.
 > "Wellington's Victory"?  I don't know how big each of the orchestras is,
    
    This is one of my faves...
    
    It sounds like there are three orchestras, but they're for sonic
    placement. Nothing a good MIDI effects box couldn't do.
    
    But you'd need a sampler and a *serious* sound reinforcement system
    for the cannons and the muskets...
    				- HBM

    Uhhh, my opinion of "Wellington's Victory" is that it's still recorded
    just to prove that even Beethoven could have a bad day.
                   
    	:-)
                                   
    -----
    
    But why use cannon and musket samples?  How about a phaser patch
    and a few klingon disruptors... :-)
             
    		:-)
    	-Bill
    

Since my question is a derivative of this topic, I'll put it here instead
of starting a new topic.  My requirement is to be able to recreate orchestral
scores of Gilbert & Sullivan light opera in real time for a community theatre
group I direct.  I expect to use some sort of "Humanizer" pad-to-MIDI-clock
box and/or maybe one of those air drum sticks for conducting the score.

The scores most often require the following instrumentation, perhaps allocated
as shown to sampled synth voices:

    Instrument  # of voices   synth voice    # synths required?
    Violin I       (2)        Violin Choir     --+
    Violin II      (2)        Violin Choir       |     
    Viola          (2)        Violin Choir       |    1
    Cello          (2)        Cello Choir        |
    Bass           (1)        Bass Choir       __+     
    Flute          (2)        Flute            --+
    Oboe           (1)        Oboe               |    1
    Bassoon        (1)        Bassoon          --+
    Clarinet       (2)        Clarinet         --+
    Horn           (2)        Horn Choir         |    1
    Cornet         (2)        Brass Choir      --+
    Trombone       (2)        Trombone Choir   --+
    Snare Drum     (1)        Snare Drum,Roll    |    1
    Bass Drum      (1)        Bass Drum        --+
    Cymbal         (1)        Crash,Choke      --+
    Tympani        (1)        Tympani          --+    1
                                                    -----
						      5

My major requirement is 16 bit sample width.  Maybe I could be talked into
a 12-bit EMAX, but I'd rather go 16.  Do I really need 5 synths?  This depth
of texture only happens in the finales, but it happens.  Maybe I could use
an HR-16 for the percussion? Does it do snare rolls?  Does it do tympani?
Tympani rolls?  If so, I *might* be able to get by with 3 synths?  The question
would most likely be the range of the clarinet voice on the 3rd synth.

What about the Kurzweil boxes or the Proteus?  Can they do this?  Maybe 1 of
those boxes could do all of this?

-jim

    I don't see why you're limiting yourself to 16 bit samplers (an aside-
    the Emax is a 8 bit companding unit).  The Roland S series are 12 bit
    linear, and sound *great*.  The EPS is 13 bit, Yamaha is 12 bit, etc.
    See note 938.6 for a decent features comparison of samplers. 

    In your application, it sounds like your limiting factor is
    simultaneous notes.  What kind of sequencer are you using?  What SGUs
    do you currently have? 

    The Proteus could certainly cover a good part of what you're trying to
    do - but who knows when you'll see one?  I've had my money down on one
    since April and haven't seen it yet. 

    The Kurzweil 1000PX is 24 voice x 16 chan polytimbral (other 1000 units
    are 20 voice), so it *could* work, but certain sounds (viz, percussion
    & horns) aren't available - at least not without adding new waveform
    memory (2 available at $400 a crack - ouch!).  You'd like the string,
    clarinet and baritone horn patches, though.  With your requirements,
    you want 25 simultaneous voices - 1 more than the 1000PX can deliver.
    Maybe a 1000PX and a 1000HX/SX (horn or string expander)? 

    As for samplers - well, the S330/S550 are 16 voice x 8 channel (?)
    polytimbral, and have great sound - but you're looking at $1500/$2300
    apiece for these units, and you'd need 2 units. 

    Another option would be a Roland U220 (list $1695), a cleaned-up U110
    with a keyboard - or the U110 itself, for that matter. 

    Another thing you might be able to get away with is using layered
    sounds - many samplers/sample players have samples of several
    instruments at once (eg, string section).  That would save you voices. 

    As for drums, the HR16 can indeed do drum rolls - but you have to
    program them (unlike the R8, which generates them automatically). 

    Given your situation, I think I'd opt for an R8 for percussion, a
    1000PX for strings/clarinet, and either an S330 sampler or another
    Kurzweil HX for the horns.  But we're talking bux here.  If time isn't
    a factor, the Proteus might be worth waiting for, too.  Are there any
    stores in the area that have their demo model available to let you play
    with it? 

    Hope this helps.

-b

    If you were driving the whole thing with Mark of the Unicorn Performer
    v3, you could feed the air stick into it directly and it would track
    your conducting.
    
    If you want real time control of dynamics, especially of balance
    between parts, you will need a bunch of continuous controllers, or
    have a guy in front of you (where the orchestra would sit) with a big
    mixer.  Opcode Vision will take controller input during playback to
    adjust various "faders", which can affect all kinds of things including
    tempo.
    
    Who is gonna sequence all those orchestral scores?  It is more than
    transcribing the sheet music in step-time.

    On the drums - the HR-16 doesn't support timpani, and its bass drums
    are not the orchestral style (they're too small and "tight").  What
    you want for this is an R-8, with the contemporary percussion card,
    which has the timpani and orchestral bass drum.  You don't want
    to use the auto-roll feature of the R-8, as for snare drum rolls
    you'll want a really tight roll that takes full advantage of the
    R-8's nuance and volume features (e.g., a typical double stroke
    roll would have the bounce stroke slightly lower in volume, the
    right hand a little stronger in volume, and the right and left
    hands with slightly different nuance).  You can also edit the head
    to snare sound balance.
    
    The R-8 can do the cymbals, snare, bass and timpani all at the same
    time, but good sounding rolls will eat up voices.
                           
    You got a lot of money or something?
    
    len.
    

    Jim:
    
    	I do (maybe attempt would be a more accurate term) a lot of
    chamber and orchestral sequencing.  I think I understand the problem
    you are trying to avoid and have faced it myself.  There is nothing
    more frustrating than hearing your music truncated due to lack of
    voices.
    
    	Having said that, one must be very careful not to overbuy for
    the one or two occaisional notes you might need in a score.  For
    example, I would not buy a ROLAND R8 just for the TIMPANI and/or
    BASS DRUM sounds.  I have heard them and though they are good, IN
    THE CONTEXT OF A MUSICAL SCORE YOU COULD DO JUST AS WELL FOR DOLLARS
    LESS.  The TIMPANI in ROLAND's D110 is very decent (not quite as
    quite as the R8's perhaps) and you also get a boatload of other
    usable sounds like FLUTES, HORNS, and STRINGS.  You should be able
    to pick up a NEW D110 for less than $600 if you shop, and could
    probably do a lot better on a used one.  
    
    	I am not disagreeing with any of the other noters, they are
    right in making the recommendations they have.  It is just that
    I have found that the trick to making realistic sounding ensemble
    works is having enough voices without breaking the budget on the
    voices or sounds you may not use more than once or twice.  If sounds
    like TIMPANI, BASS DRUM, etc. are staples in your music, and if
    fidelity is crucial, then by all means go for the best.  
    
    	Also, I have found that no ONE SGU can produce the variety of
    sounds that I need.  That is why even after vowing not to, I have
    now bought 3 of the little beasts.  Each one produces sounds in
    a different way and is better at reproducing different timbres.
    Now I can spread the polyphony across four units (if you count the
    drum machine...er Rhythm Composer (sorry ROLAND):-)) and not have
    to worry about taxing the polyphonic capability of 1 poor SGU.
    Finally, I would suggest getting as many synths and sample players
    and as few full-blown samplers as you can get away with.  Actually
    doing the "orchestration" for this kind of music is complicated
    enough without having to deal with samplers' memory and voice limi-
    tations.  However, if the ultimate in fidelity is your goal you
    may not have a choice.
    
    Clusters,
    Bill Allen

I'd like to thank everyone for their input so far.  It has helped start
me thinking about the necessary tradeoffs.  The R8 sounds like a monster
drum machine (with the ability to simulate bounce and hand dynamics? wow!).

As to the question of money, no, I'm not loaded nor have I recently come
into a large family inheritance.  But I just finished a run of the Mikado
with this group and while it was good, it could been better.  The voices
this season were wonderful.  The major problem was the orchestra manager. 
She didn't raise a full orchestra.  We had only one first violin (herself),
no second violin during the 2nd week, only one viola, no 2nd horn, and no
2nd trombone.  And she still spent our entire (albeit meager) orchestra
budget!  We had to fill out the missing parts with piano.  I raised last
season's orchestra myself, had full instrumentation (all the string parts
were doubled except the double bass), and came in under budget.  But I
worked too hard.  That's why I asked the board of directors for an
orchestra manager.  It was very frustrating.

So I'm investigating this topic for two reasons.  First, I would like the
group to do a winter show.  If they decide to do so, I think it would only
be possible if the orchestra was entirely electronic.  Second, if I get into
another situation like this summer, I would like the option of augmenting the
live orchestra with electronic instruments.

As Bill points out in .-1, sample players would help keep the price of the
ensemble down.  That is, if *I could find* a player for less than the cost
of the D550 (~$2.5K?).  I must not be informed about sample players.  The
only ones I can find are the Akai DP1000 (?) or the Oberheim, both for about
$3K, if I remember correctly.  Anyone have any info on less expensive players?

-jim

    Notice that the number of instruments desired is more than the number
    of MIDI channels.  So splits, such as  a synth playing the trumpet on
    the high part and double bass on the low part would help, and playing
    one patch for strings on one synth would help too.  Dual-bussing would
    help, anyone doing it?
    Tom

    Depends on what you mean by 'sample player', Jim.  The Kurzweil, U110,
    and Proteus are all sampler players.  Kurzweil & Proteus are "fixed",
    in that samples are blasted in ROM and are not changable. 

    The U110 can accept RAM cards, and sports "31 voice polyphony" to boot
    (some patches use 2 voices, tho).  I've heard that their going price
    will be dropping to around $800 or less soon.  FWIW.

    The DPX can play samples from many different samplers, but only has 8
    voices.  Being old technology, I wouldn't consider it. 

-b

    
    For your application, the Korg O3 Symphony module would fill the bill,
    though it is not a rack-mountable device, so it would be more difficult
    to integrate a performing rig.
    
    But these are dirt cheap (current blowout prices for the base unit are
    around $125) and by adding one sample card for about $85 to the base
    unit, you get a wide range of orchestral strings, brass and woodwinds
    to play with.  It also has both snare drum and tympani (although you
    cannot bend the pitch of the tympani...sigh).  A single O3 can do 16
    voices, so for about $600 you could have 48 voices available to play
    with.
    
    Check out note #1642 about this unit.
    
    BTW, the fidelity is not as good on this as the more expensive
    samplers mentioned, but since cost is an object...
    
    							Brian

    Stay away from the DPX. I bleeve that's the unit that won't let you
    edit the split points...
    
    Edd

>     Dual-bussing would
>     help, anyone doing it?
>     Tom

Macintosh, MTP, driving an Opcode Studio Plus Two interface

     ==> dual bus (32 MIDI channel) capable.

RE < Note 1420.22 by DYO780::SCHAFER "Brad - boycott hell." >

>    The U110 can accept RAM cards, and sports "31 voice polyphony" to boot
>    (some patches use 2 voices, tho).  I've heard that their going price
>    will be dropping to around $800 or less soon.  FWIW.

	LaSalle's (and maybe others, too?) are selling the D110 and the U110 for
$699 each right now.

	Dan

    re .25 - the Roland MC-500 with the new software can drive 32 MIDI channels
    as well.
    
    len.
    

    
    Opcode Vision has 32 instruments, assignable to a channel (1-16)
    on one of two (printer port or modem port) interfaces
    attached to the mac.
    

Notator for the ST will support up to *6* independent MIDI (out) busses with
option C-lab hardware Export and Unitor (Untior also provides SMPTE lock).
That means *96* independent channels).  I believe with Unitor you also get
2 new IN busses.

Chad

    re .29
    
    ...not forgetting C-Lab's "Human touch". That's the baton-waving
    tempo control problem solved !
    
    Jim Burke