2015-10-25 21:33 GMT+01:00 Thomas Morley <thomasmorle...@gmail.com>:
> 2015-10-25 13:00 GMT+01:00 David Kastrup <d...@gnu.org>:
>> Thomas Morley <thomasmorle...@gmail.com> writes:
>>
>>> 2015-10-25 8:34 GMT+01:00 David Kastrup <d...@gnu.org>:
>>>> Thomas Morley <thomasmorle...@gmail.com> writes:
>>>>
>>>>> Hi Bernardo,
>>>>>
>>>>> please see attached. Does it fit your needs?
>>>>
>>>> What's the essential difference to the current code?
>>>>
>>>> --
>>>> David Kastrup
>>>
>>> Current `determine-frets' from scm/translation-functions.scm checks
>>> whether the calculated fret is an integer. If not, it throws a warning
>>> and doesn't print it.
>>> At first glance this makes sense, because there are no frets for
>>> quarter-tones on a fretted instrument like guitar (in standad-tuning).
>>
>> Well, "in standard-tuning" is the point. As originally requested, the
>> idea was to have some strings tuned to a quartertone offset, and then
>> determine-frets was supposed to use those.
>
> Well, I tested the following code with default `determine-frets' and
> `my-determine-frets' from my recent post, output attached.
>
> custom-tuning = \stringTuning <eeh, a, d ges beh eeh'>
>
> mus = \relative {
> eeses'
> eeseh
> ees
> eeh
> e
> eih
> eis
> eisih
> eisis
> }
>
> tst =
> <<
> \new Staff << \clef "G_8" \mus >>
> \new TabStaff \with { stringTunings = \custom-tuning } \mus
>>>
>
> \score {
> \tst
> \header { piece = "default-determine-frets" }
> }
>
> \score {
> \tst
> \header { piece = "my-determine-frets" }
> \layout {
> \context {
> \Score
> noteToFretFunction = #my-determine-frets
> }
> \context {
> \TabStaff
> \override TabNoteHead.before-line-breaking = #my-format-tab-note-head
> }
> }
> }
>
> The second score looks nicer.
> In a local branch I ran it against our regtest, without result (maybe
> we don't have a regtest with quarter-tone-tuning, didn't check)
Ok. I found a problem: you can't bend an open string ...
>
>
> @users:
> In general, I am a professional guitarist in the classical domain
> never using tablature myself, although I'm able to read most historic
> and modern tablatures.
> (Apart from "deutsche Tabulatur" - for an image see:
> http://www.lautenmusik.net/media/lautenmusik/dt_tab1.gif )
>
> Meaning I'm not very interested in TabStaff. I'll work on it, if users
> say what they want/prefer, where are bugs etc.
> In other words: I need feedback from users, otherwise I'll focus on other
> stuff.
>
> So far only Federico and BB (as I first posted the code) reported back.
>
>
> Please test against real music!!
>
> Cheers,
> Harm
>
>>
>>> Though, ofcourse you can produce the quarter-tone pitch via bending,
>>> which then is not represented in the tab.
>>>
>>> Basically I changed it to check for (truncate fret) and removed the
>>> according warning (letting the warning for negative frets in place).
>>
>> But that would then _not_ pick quarter-tone tuned strings unless it
>> happened to find them before the others, right?
>>
>> I don't think that we can solve this satisfactorily without _scoring_
>> found combinations and picking best score. Or at least make separate
>> passes with increasingly relaxed conditions, only taking the next pass
>> when the previous one fails.
>>
>> Issue 703 suffers from the same problem: I found a viable solution
>> (patch is in the issue) that would always work as opposed to the default
>> solution, but that was not accepted by banjo players since the default
>> is the better solution _iff_ it works.
>>
>> The fallback solution as opposed to a finer-grained scoring solutions
>> would have the advantage that it is pretty much predictable, so the
>> danger that the unhelped fingering is given a different assignment from
>> one version to the next is slim. Of course, with the disadvantage that
>> a scored version will usually be better. Particularly if there is a
>> score for fingers retained from the last chord.
>>
>> --
>> David Kastrup
_______________________________________________
lilypond-user mailing list
lilypond-user@gnu.org
https://lists.gnu.org/mailman/listinfo/lilypond-user
