Robbie said:

>
> I spoke by phone with both Jody and John Roache, and they still
> kinda wonder "how is it done?", meaning the big, full synth sound.
> I could only guess that it results from lots of parallel voices
> (as in a pipe organ) and use of "chorus" or similar tremulant
> effects to get the "free phase" pipe organ effect.
>

There's no chorus, but the reverb on each channel is at 60+.   I also
spread out the image with the pan settings.   As to the big sound, it
comes from two things (I believe)

        1. octave doubling.
        2. using certain GM/GS synth voices in unnatural registers.

#1 I learned years ago arranging for strings.  If you want a bigger
sound, double at the octave, both upper and lower.

#2 I stumbled onto when I first tried creating orchestrion simulations
using only GM/GS voices on my little AWE32.  At first I used things
like "church organ" and "flute", but they just don't have the
complexity of a good organ string pipe or diapason.   However, the
"synth voices", and some of the brass do.   My favorite is the synth
"voices" played in the 2' register as string pipes.   Played in
combination with the xylophone they give a wonderful punch.

So, for instance, the "tuba" voice makes a very nice 2' reed -- not as
a solo reed, but in combination with other voices it adds a nice
resonance.

The other thing is that GM doesn't really have a good gutteral bass
reed voice -- unless again you use something in an unnatural range.   I
experimented with all the "reed" voices, and found that the Alto Sax
has the best bite in the low bass range (again on AWE32 - I don't know
what other GM synths sound like)

Although I didn't use it here, I also discovered that the "string
ensemble" played in 32' and 64' range adds a nice organ sonority.

And lastly, don't forget the GS "orchestral drum set" which is vastly
superior to the GM "rock" set for this kind of thing.

The bottom line is that I just let my ear be my guide.

> Jody has a special program he wrote which shows that at one instant
> there are 37 notes played simultaneously.  He thinks the
> SoundBlaster card in the PC is dropping notes, but the Sound Canvas
> box is handling them okay.
>

My AWE32 will drop out at about 30 voices, but makes intelligent
decisions as to which ones to kill off.   There's enough going on that
when the cancellation happens you just don't notice it.

> George, could you write up a description of your techniques for the
> Digest?  Also, I'd love to hear from you (on the Digest) what
> editing tasks you do and what programming features you'd like to
> have.  (ref msg from Mark Fontana.)  David Wasson said he'll sent
> a note soon.  I don't know any other _orchestrion_ arrangers...
>

For now I think the most important thing to know is that I don't
have a keyboard hooked up to the synth, but do everything in the
"piano roll" view.  Yup - I plonk down the notes one at a time.
I like the piano roll because I can click on a point, and then
"bleed" out the note to whatever length I need.  I started doing
this because my style is very chromatic, and Cakewalk's method for
putting in non-diatonic notes is difficult.  In fact just about
anything with a "staff" view does a poor job of entering chromatic
music.

So there isn't any post "recording" editing done.   Having said
that, most of my concentration is on (beside pitch choices) note
duration.  88 note rolls, having no "expression", leave playing with
duration and voicing as the only control you have when trying to
make sure that the "tune" is heard.   Someday, when I have an
extra $2000, I'll have a MIDI connection added to my Ampico A
and start putting in Ampico expression.

I don't use the "loud pedal" because I prefer the fine control I
can get by note bleeding.   Besides, affordable synth pianos don't
change the timbre when you sustain notes -- there's no resonant
smear, they just delay the "note off" signal.

Bleeding the notes is a large part of what I do when the piece
is all figured out.   That's where the "human" feeling really comes
from, not from velocity or random onset.

It would be very helpful to be able to see more than one "channel"
on the piano roll.  (Cakewalk doesn't allow this,  Other sequencer
programs may).  Or at the very least to be able to "scrub" the
entire ensemble while still in one piano roll view (I know there
are ways to do this, such as having all the channels represented
in the "staff" view, but they are "ways around" the problem, not
solutions).

If the target is roll stamping equipment (and thus a coarser
granularity), it would be nice to have "snap" settings that correspond
to the target granularity.   I've been experimenting
with 10 tics at 240/quarter, but am not sure it won't still cause
problems when the MIDI is converted to perf.   For the QRS roll,
I played it very safe and confined myself to 8ths, 16ths and
triplets.   No post-process "swing quantizing" because I wasn't
sure what kind of distortion would happen if the "quantized" notes
fell on a "crack".

Also "tempo" settings that correspond with feet/minute, and a "grid"
that can be realigned would make it easier to arrive at the target
with less intermediate steps/messing around.

As to other "composing" considerations, we could also talk about
that, but it would get into high composing philosophy, which could
get into some very serious bandwidth, and heated arguments.   I
haven't done enough of these to feel completely confident about
discussing compositional technique yet.   I'm still feeling my
way around.   For instance, Durrell Armstrong sent me a tape of
"National" blues rolls, and that's made me think more about
"density".

GB