Conference napalm::commusic_v1

    9 and/or 10 note polyphonic?????  What kind of keyboards do that (other
    than 10 voice Prophet 10)??

    Generally speaking, the more the better - but there is a caveat. Some
    synths steal back notes intelligently. 

    For example, most of the synths that I have (like the old OB-Xa and
    Yamaha FM stuff [I think]) do round-robin note stealing. In other
    words, osc bank 1 plays, then 2,3,4,5 ... and back to 1 again. This can
    be annoying if you're trying to play a big sustained chord with a bass
    note (or any note noticably out of the range of the "big chord") and
    suddenly play note N+1 (N=number notes polyphony). If the "out of band"
    note was the first note played, it gets stolen to play the new note,
    and the cutoff is sometimes annoying.  Of course, the larger the
    polyphonic capacity, the less of an issue this is. 

    There are other note stealing mechanisms that make a board "act" like
    it can play more notes than it actually can.  The Ensoniq boards (ESQ,
    SQ80) use a machanism where notes that are already being sounded are
    stolen - ie, if your previous chord was C2-E3-G3 and your next note is
    another G3, the old G3 is stolen and replaced with the new note, still
    leaving you with 5 notes left to play. 

    I don't know much about other mfgrs units (like Roland or Kawai). 

    Seems to me that if you do lots of piano-type sustained things, you'll
    need the extra notes (although the difference between 8 & 10 seems
    pretty moot to me). If you do lots of syncopated of pizzicato stuff,
    you may not need as many. 

    Good luck.

-b

    Alex:
    	I am certainly not an expert on this, but I don't think that
    difference between 8, 9, and 10 note polyphony would be signifi-
    cant.  For all practical purposes, on this parameter the machines
    would seem equivalent and you should indeed look to the balance
    of other features on which to make your decision.  However, I
    should also mention that the machine I use is 16 note polyphonic,
    so that you may want to consider the experience of COMMUSICers who
    are using 8 note poly machines.
    
    Bill Allen @MPO
    

    At least two of my pieces simply could not be played without
    16 notes, and not because of sustain, but because 16 notes hit at
    once 3 times a bar, a bar/second.
    Depends on your music.
    Tom

    I am just starting/learning about all this junk.  But as a new owner
    of an Ensonique EPS which is 20 note polyphonic I would vote for
    getting more than 8 or 10.  I have been using it to sequence its
    own stuff and once I have 3 or 4 instruments playing at once the
    polyphony count becomes very important.  If this is to be just your
    controller maybe it won't be so important - but for me with the
    unit being my only resource for sounds I really like the 20 note
    polyphony.
    -Tim
    

    
    
    I think that basing your decision (.0) on whether or not you get
    one or two notes of extra polyphonics is nit-picking, especially
    considering the justification (big chords).
    
    Polyphony in and of itself isn't really that big a deal these days.
    I still play mongo chords on my six-voice Korg Polysix, because
    one hand only has five fingers (at least mine do anyway). But it
    becomes more important if you are sequencing, so long as the concept
    of multi-timbralism (as opposed to polyphony) is understood too.
    
    The EPS is both multi-timbral and polyphonic. When sequencing, it
    will steal notes from other instruments to maximize voices - for
    instance, steal a note from the piano track to play the bass line,
    if necessary. The ESQ-1 will do this also. Are you going to be
    sequencing or playing more than one instrument at a time (i.e. split
    keyboard) on the board? What are the other features of the boards
    you are looking at? There are many many many options which can be
    evaluated. My last major keyboard purchase was the result of six
    weeks of concentrated shopping and comparing features, which is
    more like a union of sets problem. Polyphony was assumed. Number
    of voices in the price range, multi-timbralism, and memory capacity
    were more important though. 
    
    So what are the other features? 
    
    /pjh

    I'd go for as many voices as possible, but how many you need really
    depends on your *sound*. 
    
    	Notice how many notes you play when you hit your most complex
    chord. Even if you hit five notes on the right hand you probably
    won't hit more than three on the left. (Unless you like a muddy
    sound.)
    	I have an ESQ-1. The places where I would like to have more
    voices happen when I am using some very legatto voices, especially
    strings. When notes are stolen from a chord of strings the effect
    may not be to your liking.
    	Another area that needs more voices is when you layer your sound.
    Unless I am mistaken all synths that layer sounds do it by using
    up voices. ie: 8 voices, two layers = 4 layered voices. Do you like
    long drawn out sounds or layers? Go for more voices. 
    	Are you going to use your gear to perform or to record. In a
    recording situation you can get by with fewer voices by muting tracks
    and recording them one at a time. Of course you have pushed the
    problem out to the recorder but it may be more manageable there.
    Depends on the sound you want.
    
    	FWIW my $0.02
    	Mark
    

    If you use the sustain pedal AT ALL then you WILL hear note cut-off
    with an 8-voice or 10-voice unit.  With all the SGU's (Sound Generating
    Unit) available with 16 or more available simultaneous notes or
    voices, I couldn't recommend an ~8-voice unit as a primary axe.
    
    karl

    All the points mentioned so far are very important, but one that
    hasn't been said is "How does it feel". If this is the first of
    a series of purchases (believe me you'll get hooked) and you wiil
    end up buying rack mounts for more voices in the future then consider
    that you may be driving everything from this one board for a while,
    so touch and feel are important. 
    I suggest you also learn (if you dont already) how to read a MIDI
    implementation chart and check this out on every board you consider.
    There are a lot of facilities and parameters that may not be important
    right now but will be in the future e.g. the ability to remap
    controller numbers to match some of your potential outboard gear.
    Think long term as well as your immediate requirement. If you think
    this is unimportant then go for more voices if the unit is
    multi-timbral and you will want to start sequencing.
    Ken.
    

  reply: .all

   I think all of the responses so far are good.  I do however think that
   they are not exactly what Alex is looking for, not that it is our fault.
   Correct me if I'm mistaken but ~1-2 months ago Alex was asking about
   different 'boards in the $300-500 range, and his note is geared toward
   that range I think.  In that range, eight voices seems good I think.

	FWIW

	Chad

    I'll preface my reply with the statement that it's hard to believe
    that there aren't more important features to differentiate the
    machines.
    
    That having been said, I think the only place where 10 is signficantly
    better than 8 is if you do a lot of layering in your LIVE (non
    sequenced) playing.
    
    If you do don't do layering, it's hard to imagine anyone playing very
    many 10-note chord that won't sound like crap no matter what you play 
    it on.  That is to say, that there aren't many 10-note voicings of
    chords that I think sound good.  Much above 6, it gets far too busy
    and muddled for my tastes.
    
    However, *I* do like to play layered sounds, and I have to be very
    careful doing this on the ESQ-1 where you can only play 4 layered
    notes at a time (8 voice polyphony).   5 layered notes would be
    quite welcome, 6 would be plenty.
    
    In fact, to get around this limitation, I often make patches where
    only the upper section of the keyboard is layered so that I can play
    6 note chord voicings.
    
    Now if you're doing sequencing, the first question is, "IS the
    instrument polytimbral - can it play several different patches at
    once?".  If the answer is "no", 8 voices is plenty for sequencing.
    
    If the answer is "yes", well... 8 is not enough, but neither is 16
    either probably.  At that point, you really have to expect to have
    several SGUs.
    
    	db

Alex,

I would also recommend as many voices as possible, especially if you are
going to play a piano patch with both hands and sustain pedal, or are
using it for multitimbral sequencing.

I have had the following results with various keyboards...

Roland JX8P - 6 voices, marginally enough for once bass note plus right hand
              chord for GB work
Roland D50  - 8 voices, enough for organ and synth patches, no good piano
              patches available, would be suitable for jazz piano, Scott
	      Joplin, or no sustain usage
Kurzweil 250 - 12 voices, piano, not enough for arpeggios, but adequate for
              recording
Roland MKS20 - 16 voices, piano module, just about right for piano
Roland S550  - 16 voices, multitimbral sampler, would like to have more
               voices, but it will handle solo piano or sequencing
Ensoniq EPS  - 20 voices, multitimbral sampler, enough for piano or sequencing
Kurzweil 1000PX - 24 voices, plenty for all my piano needs

I don't think one or two extra voices is going to make any difference. Check
to see if the keyboards have last note and first note priority options for
channel stealing. What kind of keyboards have 9 and 10 voices?

dave

Gulp.  .9 was right.  The range I have been considering has been under
$500.  

The replies, as far as I can understand them, are very helpful, although
also somewhat unsettling.  

To answer some questions: It is a split keyboard that I want, and that is
the primary feature over the 9 and 10 poly keyboards.  The left side of
the split has 40 voices and the right side has 80 (if I remeber
correctly).  There is some memory, but not very sophisticated.  If I
understand layering, I don't think it has it.  I do hope to get a sustain
pedal. 

The keyboard is for performing (although not in the way you people
mean--I'm not in a band, for example), and for connection eventually via
MIDI outputs to a computer with music staff writing capability. 
Eventually is 1-2 years off, probably.  Will I get bitten by the
differences between units in the under $500 price range as far as MIDI is
concerned?

My mouth waters at what is available.  Is the right approach to choose an 
entry keyboard from which I can figure out my needs?  I figure, if I hit
it right, so much the better.  If not, then 2 or 3 years from now, you'll
be able to get what a $1500 keyboard has today for 6 or $700.  Right?

Thanks,

Alex (who feels here like he is asking who Beethoven is in the classical 
      music notes file)

    It's curious that the two halves of the split have different timbral
    repertoires.  Most split MIDI keyboards (especially keyboard/synth
    combos) allow an arbitrary mapping from any of the patches available
    to either half of the split.  Normally you'd want a bass timbre for
    the left hand and some ensemble timbre for the right hand, but why
    assume that's what you'd *always* want.
    
    len.
    

    That is bizarre.
    
    The ESQ-1/SQ-80 is particularly flexible in this area.
    
    SPLITs are part of the "program" ("patch", "sound", "timbre",
    whatever).
    
    For each program, you can pick a "split point".  The "split point"  
    can be any MIDI note, including MIDI notes which aren't on the
    the ESQ-1 keyboard.  You pick the split program and tell it which
    side of the split point it plays on.
    
    You can even add a layer to the split program.  (This featured was
    probably insisted upon by the Ensoniq marketers who were selling
    ESQ-M's ;-) ).
    
    The idea of having the split be part of the patched seemed bizarre to
    me, but it didn't take long to realize the advantages - advantages
    which I abuse regularly in my use of the ESQ-1 as a "system
    controller".
    
    	db

    Oh yeah, the bad parts of the way Ensoniq does it.
    
    The split program is recorded via it's MIDI program number (it's
    location in the program bank).
    
    This leaves you subject to a few pratfalls:
    
    	1) Putting a new program in slot n screws up the programs
           that use slot n as a split program.  And of course you
      	   have no way of knowing anyone uses the program that way.
    
    	2) When you copy a program that uses a split program off a
    	   tape/disk/etc, you are not notified that there is a split
    	   program associated with it.  Chances are your STRING/BRASS
    	   split will end up as a STRING/BONGOS split when this happens.
        
    	3) When you pull the EPROM out, references (i.e. splits and layers)
           to programs on the cartridge A (MIDI programs 41-80) and 
           cartridge B (81-120) get MODed by 40, and thus become splits
           and layers with random internal programs.
    
    Thee are some problems that can't be solved, but for some operations,
    a warning message would be greatly appreciated.
    
    	db
    	 

    The DX21 will remember split points as part of a 'performance.'
    
    You can assign polyphony 4-4, 1-7, or 7-1. The drawback is you
    assign polyphony at the patch level...
    
    Someone will do it right. Someday...
    
    Edd

Okay, I just got off the phone and here are the two keyboards:

Keyboard #1.  40 preset tones, synthesized.  Midi 8 multi-layered.  
         3 split points.  Independent voices for split keyboards although
         for some reason, half of the choices for the left hand. 
         8 note polyphonic.  Chord memory.

Keyboard #2. 30 preset tones, sampled.  Layering of two voices in any
         combination.  Registration feature allows up to (I guess) 4 
         different voices (or layered voice pairs).  Midi is 12/3 (6-4-2)
         with numbered drums.  10 note polyphonic. Real time memory.

Since this is a polyphony note, (and assuming that my bias toward a 
split keyboard at this price outweighs the slightly better sampled sound
and {I think} the layering), am I going to be seriously bitten by the
differences in MIDI here?

Thanks,

Alex

    re .15, .16 - Much weirdness abounds in this space.  The JX-10 also
    defines splits on a "by-patch" basis, but Roland's notion of patch
    is higher level than everybody else's, more like "performance" data.
    A patch consists of two sections, a lower section and an upper section,
    each of which can be assigned any "tone" (what everybody else calls
    a patch, i.e., a configuration of synth parameters that defines
    a timbre).  Splits are specified by the highest note the lower section
    plays, and the lowest note the upper section plays.  This makes
    it possible to have a "dead zone" in the middle of the keyboard,
    where neither section plays, or an overlapped middle zone where
    *both* sections play.  In addition, you can disconnect the keyboard
    from the SGUs, and define two *different* splits for the keyboard and
    the SGUs, and bind this to a patch.  (I finally found out how to
    make the JX SGUs receive on a different channel from the one its
    keyboard is transmitting on.  Unless you're a JX owner, you don't
    really wanna know.)
    
    len.
    

Six months later...

I have been on a steep learning curve.  I think I know enough now to
answer the question I asked back then.  (Much of the basically excellent
information in the previous replies was, at the time, over my head.)
I'm going to try here to summarize the replies in language that other
neophytes can understand.  See the glossary for terminology.

1. Simple definition: Polyphony basically says how many notes can be
sounded at once.  

2. Typically a synthesizer is rated at a given polyphony based on how many 
distinct sounds it can generate at once, where "distinct sound" means a 
note of a given voice.  Thus, on a multi-timbral synth, if you layer two
voices (e.g., horn and flute), the 8-note polyphonic synth will only
allow you to sound four notes at once.  I.e., with two-level layering, it
will act as though you have only 4-note polyphony.  On the Yamaha
T81z/DX11, you can layer up to 8 voices, but with 8 instruments layered, 
you can only play one note at a time (since the synth is 8-note polyphonic). 

Typically a split also reduces effective polyphony of each keyboard
section.  Thus if an instrument has 8-note polyphony, and you establish a
split, you may get, say, 4-note polyphony in the lower section and 4-note
polyphony in the upper section.  Or it could be 2-6 or some other
combination.  

3. Another critical factor in assessing polyphony is sustain pedals and
release curves.  Polyphony says how many notes can be simultaneously
sounded, including all echoes, sustains, releases, etc.  Thus, you may
only be hitting 8 notes at a time, but as you move to new positions and
old keys are still sounding, some notes will be unceremoniously terminated 
before their time in order to accept the new notes.  The abrupt ending of
a note can be very noticeable and bothersome.  

4. In general you do not play organ-like music at any level higher than
about 5 or 6 note polyphonic, but with echoes, you could easily desire
beyond 10 note polyphony.  Piano, however, often takes much more than 8,
with 16 being minimal and 24 being apparently adequate (including sustain
requirements).  

Thus the safe recommendations might be as follows:  

a. without layering or splits, 9 or 10 might be reasonable as long as you
don't plan on realistic piano sustains and the patches don't have long
release times.  

b. with layering and splits, anything below 16 is going to become
burdensome.  32 is much more likely to be a safe choice.  Thus, if you're
considering your dream synth, and have a modest budget, you're better off
getting a used synth now.  It will be a couple of years before 32-note
polyphonic synths are in the lower price ranges.  

---------------------------------------------------------------------------
RE: .18

Len, so far your JX-10 sounds like it is the weird one!  The units I have
explored have "patch" to mean a given sound quality or instrument, and
some other parameter (such as performance) to pick among patches and
define split points and layering.  I believe that if you did an EDT

    s/patch/performance   and then    s/tone/patch

we'd be speaking the same language. 

Alex

    re .19 re 1.8 - right, I agree; I didn't define Roland's terminology,
    but I have to live with it.  Often when talking to other people with
    other vendors' gear about Roland stuff I have to translate in real
    time.
    
    len.