Trill "Pause" At End
Trills, when played back at certain tempos over notes of certain durations, tend to have a "pause" at the end. I've attached a file; you can see through the pianoroll editor exactly what I mean. Several of the trills have a longer note at the end than they should. Bottom row does default trill playback, and top does baroque playback. From what I can tell, it has to deal with the note length not being properly divisible, creating this glitch.
| Attachment | Size |
|---|---|
| trill test.mscz | 11.3 KB |
Comments
Can you point specifically the notes that cause problem?
If you take a look at the example, all of them exhibit this behavior other than the eighth note, double-dotted sixteenth note, dotted sixteenth note, and sixteenth note. The latter 4 do not exhibit this behavior because they do not play.
I think I hear it. I suppose the rule is so that the trill always ends on the original note, but the code probably doesn't account for that, only adds it after the trilling is finished. What do you guys think? Can someone take a look at the code?
If that is the case it is probably done on purpose. Trilling on a 1/4 note would perhaps be to play 8 1/16 notes, as you can see the math is fine and simple there (the 1/16 note exists), and since you want to end on the orignal note you just have to play one more note afterwards. If you were to calculate this and take the last appended note into consideration you would have to play 9 1/36 notes instead. Maybe that 1/36 note indeed exists, but with stranger note lengths the math could result in really strange note lengths?? And maybe it didn't sound good that way? I am just guessing, someone should check the code, I don't understand much of that yet :)
Several details are questionable in the current ornament realisation implementation, for instance each of 'notelets' making the ornament up are shortened by the articulation of the main note (legato, non-legato, staccato, etc.); the very fact that those 'notelets' are measured is (not necessarily wrong but) certainly arguable; and so on.
IMHO, among these details, the fact that the main note can be distinctly heard at the end of the trill is not necessarily 'wrong'.
I am currently revising this part of the code while trying to (re-)implement this feature present in 1.3 and dropped in 2.x: https://musescore.org/en/node/58556 (also in #23063: No way of controlling gateTime at present in nightlies). I am trying not to alter anything of its logic to reduce the risk of regressions, but details like this are likely to change, should my Pull Request be accepted.
So, is it perhaps the case to hold issues like this back for a little while?
Oh, no issue. My only complaint is that it would sound really, really off if there are notes that actually follow the trill that normally wouldn't have a pause (think Nachschlag).
#5: indeed! But end grace notes are not played back at all currently. Should they be implemented in playback, it would be certainly the case to 'right align' (so to speak) the trill, so that any extra rounding-off is played at the beginning rather than at the end (and incidentally, I greatly miss the initial appoggiatura in the so-called 'Baroque' trill).
Yeah, grace notes should be played too, you always have the option to not play them like any of the articulations. Should probably be a song global setting for all of these... or is it already?
Well, the issue is that it also prevents any workaround from working that provides the desired playback.