Silver Apples

Talk about your microtonal "pop" music; I just saw Silver Apples last
nite, and what a hoot they are. They were actually on the scene about
1967, and made two albums (both of which I still have on vinyl); they
have now reformed, and have two new CD's out. Simeon is the leader, and
he plays with a drummer. Simeon originally played this strange handmade
rig of oscillators and such, which he built himself...last nite, he had
the homemade stuff, augmented by keyboard synths to get samples and
such. He plays killer oscillator solos over very intense grooves, but
what's so great is that his solos are way outside the 12 tone system; it
sounds like a very intuitive sense of melody and pitch, and it somehow
works and sounds very natural. I kept thinking that Hendrix would
approve, because it was very heavy in a rock sense, yet totally original
and heartfelt, without any sense of trying to be commercial. Simeon is
way into this, and they are a very interesting group. I think their
website is www.silverapples.com...check them out if you like whacked out
rock.
As far as microtonal film soundtracks, "Forbidden Planet" is a top
masterpiece, and again it's really unique and uncontrived. The Barrons
(Bebe and Louis) built circuitry for each character in the movie, and
were interested in sort of random effects; Louis said he wanted his
circuits to do things at random; he liked being surprised. I've never
heard anything quite like it. The Musician's Union made some noise
because they were not sure if it was "music," so in the credits the music
is called electronic tonalities. It is one of those ageless pieces; it
doesn't sound any different now than it did then, it doesn't sound like
part of it's "time"...Hstick

>My opinion is based upon using Ensoniq samplers since 1989. If you want to
>do a moderate amount of modulation in just intonation, 4 switchable scales
>are not enough.

That's true. But then again Ensoniq instruments respond to SysEx codes
too, noting the caveat you mentioned. And the MR series, and apparently
the ASR-X, comply with the MIDI tuning standard which, as I understand it,
has well-defined provisions for dynamic intonation.

Now that is of course assuming that you're restricting yourself to
matching octave boundaries on the keyboard with octave boundaries in the
tuning. I'll be hornswaggled if I can tell how on Earth people do this,
but the apparent truth of the matter is that the most common way to map
more than 12 steps per octave to a keyboard is mapping them linearly,
successive steps in the tuning to successive keys on the keyboard.

But anyway, if you're willing to admit that possibility, you can, for
example map a dynamic JI scheme to 19-toned subsets rather than 12-toned
subsets, you might be able to get by with 8 tables even with fairly exotic
harmony.





>Even if you use only fixed scales, if you want more than 8
>different ones then you must store those scales as separate files and then
>copy them individually to each and every sample instrument that you use.

Yes, that's unfortunate. In many ways, I personally would prefer a
single global intonation. I normally save different copies of the
otherwise same instrument separately, which is indeed annoying.

Still, with individual pitch tables for individual instruments I can
simulate a global intonation table, but with a global intonation I can't
simulate individual instrument pitch tables. So at least the solution is
general.





>The most flexible microtonal architecture I've seen in a commercial
>synthesizer is the Yamaha TX802. You can define up to 64 keyboard scales on
>a cartridge, and then assign each part to any of the scales.

That sounds interesting. Are those "keyboard scales" completely
arbitrary though? Can any MIDI note number be tuned to any arbitrary pitch
irrelevant of what you tune any of the other MIDI notes to (within the
range and resolution of the sound-generator that is)?

If so, then I'd say you're right that they've got Ensoniq beat, and I
think I have some product investigation to do! If not, then I'd have to
see what are the restrictions. If for example, all Cs (for example) must
be exact octaves apart, or if the step size between all pairs of adjacent
keys has to be the same or of limited resolution, I'm not terribly
interested. (Those are, as I understand it anyway, the limitations that
the K2500 puts on you. Although they apparently have something or other to
at least partly get around that; I think it's called a "function".)

Oh, is the TX802 still a current-produced product?

Roland/Yamaha tuning table resolutions are extremely poor (aren't
they)Seems like that outweighs any ease of use advantage they
have.Ensoniq : 256 increments per half-step,@0.4 cents !

Someone recently mentioned the Kyma,I think.That has the equivalent of
@10,062.461 increments per half-step, @0.01 cent/0.0026 Hz
resolution.But! ->Cost: $4400 to @$8365 ,:-)

If only synth manufacturers implemented the MTS to the letter.

Later,
Drew

> From: mr88cet@texas.net (Gary Morrison)
>
...And the MR series, and apparently
the ASR-X, comply with the MIDI tuning standard which, as I
understand it,
has well-defined provisions for dynamic intonation.

I recommend the Ensoniq ASR-X over the ASR-10 because of improvements other
than the microtonal features. Currently it is the best (keyboardless)
workstation for full keyboard scales, with the Korg Trinity series as a
strong second choice. But I recall someone mentioning that the dynamic
intonation feature was not implemented in the MR Rack.

> Now that is of course assuming that you're restricting yourself to
> matching octave boundaries on the keyboard with octave boundaries in the
> tuning. I'll be hornswaggled if I can tell how on Earth people do this,
> but the apparent truth of the matter is that the most common way to map
> more than 12 steps per octave to a keyboard is mapping them linearly,
> successive steps in the tuning to successive keys on the keyboard.
>
> But anyway, if you're willing to admit that possibility, you can, for
> example map a dynamic JI scheme to 19-toned subsets rather than 12-toned
> subsets, you might be able to get by with 8 tables even with fairly exotic
> harmony.
>
You're forgetting the 4 pitch table selection limit. Anyway, since I don't
have a generalized keyboard it is easier for me to match the 12 note per
octave limit of my keyboard to my just intonation scales, and then switch
between scales if I need some new intervals. Matching octave boundaries for
scales over 12 notes would make things easier, in my opinion. I was never
able to visualize a linear mapping properly. The limit here is our
interface to the piano keyboard, not the intonation tables.

> >The most flexible microtonal architecture I've seen in a commercial
> >synthesizer is the Yamaha TX802. You can define up to 64 keyboard scales
> on
> >a cartridge, and then assign each part to any of the scales.
>
> That sounds interesting. Are those "keyboard scales" completely
> arbitrary though?
>
> If so, then I'd say you're right that they've got Ensoniq beat, and I
> think I have some product investigation to do! If not, then I'd have to
> see what are the restrictions. If for example, all Cs (for example) must
> be exact octaves apart, or if the step size between all pairs of adjacent
> keys has to be the same or of limited resolution, I'm not terribly
> interested. (Those are, as I understand it anyway, the limitations that
> the K2500 puts on you. Although they apparently have something or other
> to
> at least partly get around that; I think it's called a "function".)
>
> Oh, is the TX802 still a current-produced product?
>
Each note is tunable over the full MIDI range (uses same programs and pitch
tables as DX7IIFD). It's FM timbres make a good counterpoint to today's
wavetable synthesis. The Yamaha TX802 has been out of production for quite
some time, but you can pick up used units fairly easily for $300-$400. You
need a memory cartridge to do the 64 tuning tables, however, so make sure it
comes with one as those are difficult to find these days.

K2000/K2500 octave-based intonation tables allow retuning over the entire
MIDI range for each note. Even the tonic can be retuned, which is why it
makes modulation in just intonation so easy. nTET and nCET scales can be
created at the program level by adjusting a few parameters, and are
unlimited in step size. Resolution is 1 cent or better if done correctly
(see my web site).

> From: sethares@eceserv0.ece.wisc.edu (William Sethares)
>
> >The other problem is that the Ensoniq pitch table sysex codes were
> never fixed through 3 generations of samplers (EPS, EPS-16 Plus and
> ASR-10).
>
> This is somewhat misleading. It is possible to store the Ensoniq
> tuning tables in your computer. I know since I wrote a program
> several years ago to do just this. (Unfortunately, it was on an Atari
> computer, so it will be of limited usefullness to most). Thus - the
> problem is not with the "Ensoniq pitch table sysex codes", but rather
> that the authors of the software did not bother to implement the
> Ensoniq pitch tables. As far as I know, though, you cannot update
> them except by sending a whole pitch table, that is, you cnnot retune
> "one note" at a time.
>
You are correct. I was mistaken on the Ensoniq SysEx codes -- chalk it up
to faulty memory. Only the individual note SysEx codes (realtime updates)
are broken, not the pitch table SysEx dump. In fact, 5-6 years ago I worked
on a patch editing template for EPS and EPS-16 Plus Pitch Tables for Sound
Quest's "MIDIQuest" generic MIDI patch editor, available for most computer
platforms. It shows all the pitch adjustments at once. But it updates the
table by dumping the entire thing, rather than just changing each parameter.
As I recall it only worked on the first pitch table, but I could probably
fix that if anyone is interested in it. I also had working instrument and
layer templates. The wavesample template was never completed due to several
obstacles, the worst being that wavesample "all" editing SysEx did not work.
These SysEx problems were eventually acknowledged by Ensoniq, after much
arm-twisting on my part.

> From: Xou Oxno
>
> I own a 05R/W. 12 note scales, +/- 99 cents range per note.
> One user tuning. I'm 99% sure the whole family has those specs.
> I know the Trinity does. I think the X5DR has twice as many voices - 64.
>
Older Korg equipment has the +-50 cents limitation. I'm not sure where the
cutoff date/line is. I'm dependent on user feedback to keep the Microtonal
Synthesis Web Site up to date since I cannot possibly get access to all
those instruments.

From: "Andrew L. Kaye"

I'm actually looking for a module that is
not limited to 12-note scales. Thanks, Andrew Kaye

In that case I recommend Ensoniq's MR Rack or ASR-X, Korg's Trinity series
or an Emu module. If price is an issue then used Yamaha FM synthesizers and
Ensoniq samplers can usually be found for less than $600, but then you give
up the internal effects buses of the current units.

> From: mr88cet@texas.net (Gary Morrison)
>
> I don't claim to have used much SysEx stuff to my Ensoniq gear, but I
> seem to recall that it implements a "virtual keypad" interface, meaning
> that SysEx messages simulate fingers hitting keys of the keypad. If so,
> then in concept I would think that you could retune some note by virtual
> keystrokes, then extrapolate (duplicate in all octaves, or whatever
> interval) it.
>
The EPS SysEx virtual keypad interface had problems that kept it from being
useful. As I recall I tried to use it to select pitch tables to get past
the 4 pitch table selection limit, and it kept switching back to the
instrument parameter page instead of the layer page that I needed to access.

John Loffink
jloffink@pdq.net

>You're forgetting the 4 pitch table selection limit.

There are ways around that of course. Somebody suggested using what
might be called "instantaneous crossfades" between layers with the
mod-wheel.




>Matching octave boundaries for
>scales over 12 notes would make things easier, in my opinion.

I'll tell you sir, I'm with you on this one. But I think you'll find
that - mind-bogglingly if you ask me - we're pretty darned near alone.
Among the people who map(ped) tunings linearly, mismatching octave
boundaries in the tuning with octave boundaries on the keyboard, are,
Easley Blackwood, Ivor Darreg, Jonathan Glasier, Harry Partch (on the
Chromelodeon), Brian McLaren, Bill Sethares, and my K2500-playing friend
Bill Meadows. And speaking of whom...




>(K2500): nTET and nCET scales can be
>created at the program level by adjusting a few parameters, and are
>unlimited in step size. Resolution is 1 cent or better if done correctly

Perhaps you can provide some insight for Bill Meadows here: He wants to
map 19TET linearly across the keyboard, which means that you have to set
the step size between keys to about 63.1579 cents. Keymaps are
12-tone-per-octave-based, so you can't use them. The other available
tuning facility was to state a step size between keys, consistent across
all of the keyboard, but the resolution of the step-size between keys was 2
cents. Obviously then setting that to 64 would make the octave sharp by
about 16 cents.

He and one of the techies at Kurzweil concluded initially that the K2500
had no means of doing that, except by assigning a single (I don't know the
Kurzweil terminology, but the Ensoniq-equivalent term would be
"wavesample") to each individual key and setting its root pitch so that it
works out right. He says that he later heard of something I think he
called a "function" that conceptually should make it possible.

Obviously though a general, arbitrary pitch table would be a lot easier.
Do you know of an easier way to do that?





>It's FM timbres make a good counterpoint to today's
>wavetable synthesis. The Yamaha TX802 has been out of production for quite
>some time

Does it by chance allow resonant filters to manipulate FM-generated sounds?

> From: mr88cet@texas.net (Gary Morrison)
>
> >You're forgetting the 4 pitch table selection limit.
>
> There are ways around that of course. Somebody suggested using what
> might be called "instantaneous crossfades" between layers with the
> mod-wheel.
>
But crossfading between 2 layers will halve the already limited polyphony
of the Ensoniq samplers, because even if the layer is silent it will still
rob one note of polyphony. So it is theoretically possible, but not very
practical. Velocity switching would not have this problem, but control is
an issue. There just is not an easy solution. Believe me, I spent years
trying to workaround the limitations of the Ensoniq sampler architecture
before I finally moved on.
>
>
> >Matching octave boundaries for
> >scales over 12 notes would make things easier, in my opinion.
>
> I'll tell you sir, I'm with you on this one. But I think you'll find
> that - mind-bogglingly if you ask me - we're pretty darned near alone.
> Among the people who map(ped) tunings linearly, mismatching octave
> boundaries in the tuning with octave boundaries on the keyboard, are,
> Easley Blackwood, Ivor Darreg, Jonathan Glasier, Harry Partch (on the
> Chromelodeon), Brian McLaren, Bill Sethares, and my K2500-playing friend
> Bill Meadows. And speaking of whom...
>
If they can do it, all the power to them.
>
>
> >(K2500): nTET and nCET scales can be
> >created at the program level by adjusting a few parameters, and are
> >unlimited in step size. Resolution is 1 cent or better if done
correctly
>
> Perhaps you can provide some insight for Bill Meadows here: He wants
to
> map 19TET linearly across the keyboard, which means that you have to set
> the step size between keys to about 63.1579 cents. Keymaps are
> 12-tone-per-octave-based, so you can't use them. The other available
> tuning facility was to state a step size between keys, consistent across
> all of the keyboard, but the resolution of the step-size between keys was
2
> cents. Obviously then setting that to 64 would make the octave sharp by
> about 16 cents.
>
> He and one of the techies at Kurzweil concluded initially that the
K2500
> had no means of doing that, except by assigning a single (I don't know
the
> Kurzweil terminology, but the Ensoniq-equivalent term would be
> "wavesample") to each individual key and setting its root pitch so that
it
> works out right. He says that he later heard of something I think he
> called a "function" that conceptually should make it possible.
>
> Obviously though a general, arbitrary pitch table would be a lot
easier.
> Do you know of an easier way to do that?
>
Achieving nTET and nCET scales of any degree on K2000/K2500s is explained
in detail at my web site:
http://freeweb.pdq.net/jloffink/
All that's required is to edit 3-6 parameters per program layer. Editing at
the sample roots or keymap level is not necessary.
>
>
>
> >The Yamaha TX802 has been out of production for quite
> >some time
>
> Does it by chance allow resonant filters to manipulate FM-generated
sounds?
>
No filters at all, just FM. Yamaha did not implement resonant filters with
FM until the SY77 and SY99, which are limited to 2 keyboard scales.

John Loffink
jloffink@pdq.net

>> >(K2500): nTET and nCET scales can be
>> >created at the program level by adjusting a few parameters
>> Perhaps you can provide some insight for Bill Meadows here:
>Achieving nTET and nCET scales of any degree on K2000/K2500s is explained
>in detail at my web site:

I'll pass that on to him.

Thanks for the ideas.

>>> >(K2500): nTET and nCET scales can be
>>> >created at the program level by adjusting a few parameters
>>> Perhaps you can provide some insight for Bill Meadows here:
>>Achieving nTET and nCET scales of any degree on K2000/K2500s is explained
>>in detail at my web site:

Thanks again for the info.

Best I can tell from your website, and from what I've heard, the K2x00
does not make possible my 88CET keyboard mapping, wherein the gap between
adjacent keys is 88 cents except for that between G#/Ab and A, wherein its
zero. Since it's not completely equally tempered (because of the skipped
G#/Ab key), and since it doesn't repeat in octaves, it sounds like it
doesn't fit either provision.)

Is that correct?

> From: mr88cet@texas.net (Gary Morrison)
>
> Best I can tell from your website, and from what I've heard, the K2x00
> does not make possible my 88CET keyboard mapping, wherein the gap between
> adjacent keys is 88 cents except for that between G#/Ab and A, wherein its
> zero. Since it's not completely equally tempered (because of the skipped
> G#/Ab key), and since it doesn't repeat in octaves, it sounds like it
> doesn't fit either provision.)
>
> Is that correct?
>
Nonoctave equal temperaments are not a problem. I gave examples of Carlos
alpha, beta and gamma and Bohlen-Pierce at my web site. Because of the
skipped key, however, I'm not sure you could accomplish this scale on a
K2000/K2500. It might be possible to adjust the pitch every nth key using a
quantize function, but I'd need a better understanding of your scale to do
this, such as a mapping from MIDI note number to cents for the entire
keyboard. I'm not sure where the successive gaps will occur since your
scale is non-octave repeating and the G#/Ab of the first octave is not the
same as the other octaves.

> From: david first
>
> As someone who has been de-tuning digital synthesizers - Casios
> (CZ's&VZ's),
> Yamahas (DX7II & TG77), and Kurzweil (K2000) - for twelve years, I am
> constantly surprised by what is considered acceptable pitch resolution for
> just tunings. Its been my experience that even the simplest of pitch
> arrays
> can be frustratingly arbitrary on any of the above named instruments. Of
> course, this would clearly be the case on the Casios and Yamahas which all
> had
> resolutions larger than one cent, but what finally opened my eyes was my
> initial experience with the k2000 which has a fine tuning resolution of
> one
> cent. The Kurzweil had less pure intervals available than the others! Upon
> reflection I realized that due to the larger than one cent tuning res. of
> the
> Casios (CZ - 1.667 ct res./VZ - 1.5625 ct res.) & the Yamahas (1.171875 ct
> res.), at least once in a while one could stumble upon a relationship that
> came somewhat close to a purely tuned inverval. On the other hand, the
> K2000,
> with its 1200 parts per octave was in essence a type of "equal
> temperament"
> with no possiblity of even accidentally coming close to the decimals
> needed
> for seriously just tunings.
>
> Since this experience (in 1992) I had the realization that the only way
> to
> achieve pure ratio derived tunings on a digital synth was to make my own
> samples at the desired frequencies and import them into my K2000. Perhaps
> some
> acceptable fudging can be done with the above instruments if one is
> composing
> percussive/pianistic types of music, but make no mistake - one is still
> only
> approximating just intonation when using ANY commercial hardware synth's
> tuning table.
>
So what is the minimum pitch resolution that is acceptable? Is this for
music comprised mostly of long sustained harmonies?
>
John Loffink
jloffink@pdq.net