Tuplets

• Aug 1, 2012 - 10:28

In this thread there's a discussion about importing midi containing triplets. I thought I'd create a new thread for this suggestion.

I haven't found a way how to change note values to tuplets. In museScore, one seems to be able to write new tuplets but not change existing notes into tuplets. If you manage to import midi files containing say triplets, MuseScore rewrites them to nearest non-triplet values. One should be able to change them into triplets after that.

I mean, you can mark a quaver, press 4 and you get a crotchet plus a crotchet rest. How do you do something similar when you want tuplets. Here...
triplets.png
I'd like to be able to change the upper line into the lower line by selecting the accented notes and turning them into 1/8 triplets. The upper line is how it looks, when a simple midi file is imported to MuseScore with 1/64 precision. MuseScore doesn't seem to be able to detect triplets in midi files, but that's an issue of the other thread. This is about turning existing notes into triplets (or other tuplets).

F.i. in 3/4, 3/8 or 6/8 meter, you might want to change two dotted notes into a duplet. They kind of mean the same, if you only think of durations, but two dotted notes usually mean more accentuated, syncopated rhythm, while duplets take the accent away from the second note.

Attachment Size
triplets.png 30.89 KB

Comments

That would indeed be a very useful thing.

I have in the past ended up re-entering notes when I have needed to change existing rhythms into triplets.

If of course you are importing a MIDI file this would be a godsend as currently MuseScore is unable to deduce triplets from a MIDI file.

I think it's time someone worked out an algorithm for this. Finale has a button you press to tell it that the MIDI file contains triplets, but it is still inclined to get them wrong!

In reply to by ChurchOrganist

And for good reason ... a MIDI file contains no informations about the triplets (or tuplets). The editor (Final, MuseScore or others) should "divine" based on the length of n notes cover exactly the duration of a figure of note. Not really easy to do and I congratulate in advance those who will find the right algorithm.

In reply to by [DELETED] 15468

I would like to agree. Importing MIDI in order to create notation is something I'm sure a lot of MuseScore users do. I have had to do some professional arrangements in the past, and all I had to work with was a MIDI music file.

I'm not sure if Sibelius and/or Finale can do this, though. My understanding of MIDI is that it alters the tuplet to duration... But I might be wrong.

In reply to by jotti

Well actually it does.

Note durations are stored as durations based on the time resolution stored in the header chunk, and these are usually defined as ticks per quarter note, and also contain the tempo, as quarter notes per minute.

Each note has a relative duration parameter according to this stored time resolution. So, if the resolution is 96 ticks per quarter note for example, it is easy for software to read these durationa and related them to the standard rhythm types of 1/2, 1/4, 1/8 etc.

The problem with a triplet is that it is 3 notes in the space of two, each of which will have a tick value which does not correlate to any of the basic rhythm types.

So in order for a notation program to read a triplet from a MIDI file it has to know how to calculate this odd number of ticks and translate it into a triplet. In FInale there is a button to press in the MIDI file input dialogue if there are a large number of triplets in the file. It will then endeavour to find them - but IME usually gets at least 25% wrong.

Hope you understand all this :)

In reply to by ChurchOrganist

If there are 96 ticks per quarter note and you have two 8th notes, they occure at tick 0 and 48. If you have a triplet, they occure at 0, 32 and 64. Four 16th notes would be at 0, 24, 42 and 72. This is of course the exact values. If you've recorded it in real time, you have some inaccuracy there. Also the lengths of the notes is just a number of ticks, not any properties corresponding to quarter notes, 8th notes etc. Note down may occure at tick 0 and note up at tick 47. The note up may of course occure at 31. That alone doesn't make it a triplet note. It might as well be just a staccato quarter note.

When you say it's easy for software, you mean it's easy for the programmer, right? It's easy to only divide by 2 consecutively. It's more tricky to mind the other divisions inbetween. A triplet 8th note is kind of a 12th note. A two beat quintuplet (is that how it's called?) note would be a 10th note. The reason why MuseScore can't find triplets is not because midi doesn't contain triplet info. The reason is that no coder has got into that yet. It's way more trickier than just finding the 8th, 16th and 32nd notes. But midi doesn't contain those either!

In reply to by jotti

Of course music recorded in real-time presents a whole rake of problems for a notation program. It will be certain that very few of the note durations will be exact - human players put in subtle nuances of timing which the sequencer will record faithfully, but which will result in a jumbled mess of tied notes when imported into a notation program.

Finale is certainly no better at dealing with this than MuseScore.

So, for this reason it is a must to examine a MIDI file in a sequencer before you attempt to import it into MuseScore, and, if the "notes" do not line up with the durations grid, apply appropriate quantisation to the file and resave it before attempting import to MuseScore (or any other notaion program for that matter)

When I said software, yes, I did mean programmer :) Dividing by 2 is natural for computers after all :)

In actual fact given the resolution of 96 tpq then it is relatively easy to find triplets, as you demonstrated, but the problem is that the resolution can be set to any value in most sequencers, so it is possible for a tick resolution which is not divisble by 3 to be set at the beginning of the file.

MuseScore's internal tpq is 720 I believe, which is a bit coarse by my standards - I always have tpq set at the highest it will go in Sonar 3 which is 960 - it can make all the difference when trying to render nuances of performance.

(Update: I've just established that MuseScore's tpq is 480}

In reply to by ChurchOrganist

Yes. Quantizing in a sequencer pays off. But I'm afraid the sequencers usually only adjust the note down event, leaving the note length untouched. This certainly would be a task for the sequencer, not for MuseScore. Of course, if any freeware (even better, an open source) sequencer could handle the quantizing perfect for any standard non-tuplet as well as tuplet notes, both note down and note duration, developing of MuseScore's midi import could be based on that. I mean, you could have some settings in MuseScore, where you define same values as you define in the sequencer for the quantization. Threshold like values. Like when to interprete the events either like 8th note plus 8th rest or just like a quarter note. The logic in MuseScore wouldn't have to deal with the inaccuracies in real time recordings. I bet most of the users that would benefit of this, wouldn't mind having to use an extra application.

I haven't used Sibelius very much, but I remember it has lots of overkill things involved with the real time recording. I had troubles getting simple quarter notes. They became 8th notes and 8th rests instead. Or they got staccato dots or tenuto marks and whatnot. And it even adjusted the tempo according to my playing, which really screwed everything up, since I had a small midi lag there. Sure it was a question of improper settings, but anyway.

In reply to by [DELETED] 15468

On my computer with win xp there is still a sequencer program called Cakewalk dated January, 1995.
As a test I just entered some measures with triplets in MuseScore. This I saved as MIDI file. Then I used both, MuseScore and Cakewalk to open this MIDI file and display the staff with default settings. See the enclosed screenshots for the results and judge yourself.
But this is a little bit of topic. The original request was not to identify triplets but change a series of three notes to a triplet. This, I agree, would indeed be very helpful.

Do you still have an unanswered question? Please log in first to post your question.