Conference napalm::commusic_v1

    Strictly speaking, there are only three forms of synthesis:
    The good,
    The bad,
    and 
    The indifferent, (which is, by definition, exactly the same as the
    other two.
    
    	Eir�kur
    

    There is no "best" or "worst" form of synthesis, each form has it's
    own strengths and weaknesses, often depending more on the tastes and
    skills of the programmer than to any objective criteria.
    
    I'm partial to FM, but only because I cut my teeth on it and know
    it pretty well.
    
    Edd

    
    >I'm partial to FM, but only because I cut my teeth on it and know
    >it pretty well.
    
    As someone who grew up on old modular gear, I *hate* FM.  I feel more
    comfortable with filters, oscillators, envelope generators etc...
    
    In general, like having lots of modules, operators, effects or whatever
    and being able to connect them arbitrarily.  Most contemporary synths
    lack the programmability I crave.
    
    Oh, one final thing, I *HATE* new names for things that have been
    around for a while.  KORG has "Timber Envelope Generators" I'd sooner
    call them filters and be on with it...
    
    My $.02
    
    -Ed

    
    Re: .0
    
    Oh, to me terms like Additive, Subtractive and Nonlinear are better
    terminologies for synthesis.  The ones you mention are really brand
    names.  Additive is sort of like LA (Adding harmonics) Subtractive is
    classic analog (my favorite) where you filter out harmonics from a rich
    source signal. Nonlinear is like FM where you can modulate one simple
    waveform with another and get some harmonically rich signals.  PCM is
    glorified sampling. Actually sampling + data reduction + processing.  It
    offers the ability to manipulate natural sounds rather nicely.  I don't
    consider it synthesis because the source is "organic", not synthetic.
    
    -Ed

    For instance, A M1 has better sound than a K1.. Is it because of the
    type of synthesis used?
    

    If you investigated, you'd find that the K1 and the M1 have essentially
    the same architecture, but it is drastically cut back in the K1.  I
    actually find the K1 to have a very good "sound" quality--I suspect
    that the difference that you hear is in the on-board samples.  There
    are many more and longer samples in an M1.
    
    	Eirikur
    

    I'd hate to call L/A additive, it's purely *subtractive* to my way of
    thinking.  I really get a kick out of all these manufacturers coming up
    with crazy names for the same synthesis process.  
    
    Isn't there already a primer note on synthesis types, Mr. Moderator?
    
    				
    							Brian

RE: Edd's comment...

	I was always an analog enthusiast (or should I say - subtractive?).  
Lately, though, I have been noticing how incredibly useful much of the stuff I
have in my TX81Z have become.  In point of fact, if I had to live with just one
module, it'd probably have to be the TX81Z (a year ago, I'd have said an analog 
module, fersure!).

	FM synthesis, in the hands of a good programmer (or if you're not one of 
those, you have access to a bunch of great programs from someone who is) can be
incredibly useful.  The biggest hangup I've found with FM synthesis (and I 
should specify 4-operator FM synthesis here) is that its difficult to get a 
patch to be timbrally dynamic at a usable volume level.  It just seems to be
the way it was designed.  I've heard discussion of this from my favorite FM
programmer (James Fellows of Angel City Audio).  He basically says this is where
4-op FM synthesis loses its simplicity and becomes a matter of intuition and 
experimentation.  It seems that there are a number of ways to get the same sound 
in FM, but there seems to be better ways to make patches dynamic than others.

	Any comment on this, Edd?

	Dan

    I remember the K1 had PCM.. but what were the others waveforms on the
    K! called
    

The synthesis method used in the K1 is very similar to that used in the Roland
L/A machines, the Korg M1, the Yamaha SY22, even the old Ensoniq SQ-80.  All of
these machines use wavetables which store digitized waveforms.  These are not
necessarily samples, in the case of the SQ-80, for instance, most are not, but
the current trend is towards samples in the wavetable. The term PCM refers to
the method Kawai used to digitize its wavetables.

The "innovation" of L/A was that Roland added *attack* samples to the usual
pallette of sustaining waves that synthesizers had always had.  That is, some
of the waves sound as long as you hold a key, while the attacks sound once when
you hit the key and disappear.  What L/A does is allow you to layer attacks
with sustaining waves so you can do things like graft a bell strike onto a
sustained sample of voices going "ahhh", a sound you hear all over the place
these days.  This is exactly how the other synths mentioned above work. 
Ensoniq called it "cross wave synthesis" (i.e. cross fading from one waveform
to another), Yamaha calls it vector synthesis in the SY22, because you can
control the mix of the waveforms in real time with a joystick (you can do this
on a D-50 or the keyboard version of the K1, too), and who knows why Korg calls
it AI synthesis, but they all do the same thing.  You may prefer one over
another based on what waveforms are available in the unit, but conceptually
they are very similar.    

This idea of layering complex waves to create sounds has gained popularity
because it's easy for a beginner to build sounds this way.  Want a piano?  Use
the piano wave and graft on the hammer attack sample.  That's easier than
trying to make a piano out of square waves using filters and envelope
generators.  It's not actually additive synthesis, since that term is used to
describe systems where you build sounds up from their harmonic components. 
It's really no different than the subtractive synthesis that's been with uas
since the early analog synths.  You take a wave, feed it through a filter to
adjust harmonic content and then give it an amplitude envelope.

The K1 and some other synths depart from the subtractive model somewhat in that
they throw away the filters.  This is a cost measure, as there is no sonic gain
from not having filters.  Many early "digital" synths were actually hybrids,
like the Ensoniq ESQ-1 which had digital waveform generation but retained
analog filters.  Since filtering in the digital domain is compute-intensive,
when Roland invented L/A, they only allowed use of filters with "synth"
waveforms (square, pulse) but not with sampled waveforms.  The idea, I
believe, was that filters were not needed, because the harmonic content was
"preset" in the samples.  Of course, if you thought that the sample was too
bright, too bad.  Also, there was no way to modulate the timbre with velocity
(i.e. get brighter the harder you hit the keys) or with an envelope (i.e.
classic analog filter sweeps).  So more recent machines have begun to offer
digital filters, some even offering resonance.  This is due mostly to
microprocessors getting more powerful and cheaper, so the designers are putting
them back in.


							Brian