Re: Turkish makam using regular.ly

[Hans Åberg]

> On 25 Oct 2018, at 06:02, Adam Good <address@hidden> wrote:
> 
> On Tue, Oct 23, 2018 at 9:42 AM Hans Åberg <address@hidden> wrote:
>> 
>> > On 23 Oct 2018, at 04:56, Adam Good <address@hidden> wrote:
>> > 
>> Does it sound lower in standard Turkish performance, that is, which does not 
>> refer to a MIDI? —
> 
> I just listened to some midi output. Assuming I did this correctly,
> #(define-public EKSIK-IKI 5/20)
> #(define-public EKSIK-UC 3/10)
> 
> ...which would be 2.5k vs. 3k as far as I know, do sound strikingly different 
> from one another, surprisingly so. That's the mechanical part. In practice of 
> Turkish performance by the masters, they simply do what ussak needs to do 
> according to their ears and lots of that has to do with glissando. Fairly 
> impossible to notate and reproduce via synthesis.

Anyway, such values risks may not work with transposition, so an alternative 
might be using a higher ET (see below). For example, Ozan Yarman has an idea to 
use what essentially is 159 = 3*53, I think. He has a video here.
  https://www.youtube.com/watch?v=y4XuVBuGa08

>> What does the Makam Uşşak MIDI sound now that you have proper E53 without 
>> that addition?
> 
> At least closer to what makam Ussak needs by the availability of 3 different 
> segah pitches but of course not perfect. But this isn't a concern to me.

In LilyPond, pitches are tied to the notation, so one would have to use a 
higher ET to ensure transposition. I think that is in effect what you already 
do: all numbers are multiple of 6, so you can experiment with intermediate 
values until the MIDI sounds right. For example, -15/53 is one half comma 
between -18/53 and -12/53. Then add suitable glyphs to work with transposition.

From what I can see, Ussak is the same Arabic Bayati, and in the latter the 
average is two commas. So perhaps there is an influence, there, the Turkish 
microtonal values have drifted so as to become larger.

>> Also note that the regular.ly file retunes around E12 C4, so its A4 will be 
>> slightly higher than 440 Hz.
> 
> True and weird!

Indeed, but in fact the same in the microtonal program Scala!

> Any idea where that would put A4?

In terms of logarithmic fractions of an octave rational interval 2, the E12 C4 
is 9/12 = 3/4 below the E12 A4 at 440 Hz, and then the E53 A4 is 40/53 above 
that C4, so it is 40/53 - 3/4 = 1/(4*53) =  of an octave. This is 1200/212 
5.660 cents, or the pitch 400*2^(1/212) = 441.441 Hz.

>> Turkish notation does not have glyphs for all E53 notes, so you might fill 
>> them in with something, to get that instead of an error.
> 
> I'm fairly reluctant to add glyphs and in fact already added one for -18/53 
> because makam Huzzam needed it for the key signature. I'll come up with 
> better fixes by way of errors rather than shoot in the dark.

I thought of it as a debugging tool, so one can see in the score if something 
went wrong (see below).

> Also added -15/53 for the ussak si

You might add the new Helmholtz-Ellis accidentals, which will kick in LilyPond 
2.21. It seems nothing happens if there is not glyph available for the glyph 
name, so adding them now. From the names that Torsten Hämmerle gave, I get:

% For LilyPond 2.21:
NewEqualFiftythreeGlyphs = #`(
  (-72/53 . "accidentals.flatflat.2down")
  (-66/53 . "accidentals.flatflat.1down")
  (-60/53 . "accidentals.flatflat")
  (-54/53 . "accidentals.flatflat.1up")
  (-48/53 . "accidentals.flatflat.2up")

  (-42/53 . "accidentals.flat.2down")
  (-36/53 . "accidentals.flat.1down")
  (-30/53 . "accidentals.flat")
  (-24/53 . "accidentals.flat.1up")
  (-18/53 . "accidentals.flat.2up")

  (-12/53 . "accidentals.natural.2down")
  (-6/53  . "accidentals.natural.1down")
  (0      . "accidentals.natural")
  (6/53   . "accidentals.natural.1up")
  (12/53  . "accidentals.natural.2up")

  (18/53  . "accidentals.sharp.2down")
  (24/53  . "accidentals.sharp.1down")
  (30/53  . "accidentals.sharp")
  (36/53  . "accidentals.sharp.1up")
  (42/53  . "accidentals.sharp.2up")

  (48/53  . "accidentals.doublesharp.2down")
  (54/53  . "accidentals.doublesharp.1down")
  (60/53  . "accidentals.doublesharp")
  (66/53  . "accidentals.doublesharp.1up")
  (72/53  . "accidentals.doublesharp.2up")
)


>> 3. Many of the not so obvious \override KeySignature #'padding-pairs = #'(
>> > are incomplete but these issues come up during weird transpositions. A bit
>> > time consuming to test.
>> 
>> The standard key signatures should now work, if you just find a suitable 
>> keyAlterationOrder. If you can, check with the unstable version.
> 
> Between the stable vs. dev versions I'm not finding inconsistent behavior of 
> key sig padding.
> 
> Check the attached for makamGlyphs definitions, note blanks as unneeded 
> placeholders for future errors. Is that acceptable?

It seems that LilyPond merely inserts no glyph and does not issue a diagnostic, 
so it makes no difference. So  even if you do not want to have that in the 
final version, you might add some valid glyph names for debugging purposes.