Slur and phrase end notes should play at 90% length

• Jun 10, 2012 - 07:10
S5 - Suggestion

What I mean is, a staccato note (which is, there is no slur on it, or it is the last note of a slur, and no legato before it, or a staccato before it, and there is not tenudo or staccato dot above that note) should play by its 90% length. I think this can make the playback more natural.

I know that Overture software behaves like this. And Overture has a playback option window for further customize, which can be shown when you right click a note and choose playback settings.


Are you talking about an instrument in particular or all of them ? In particular Piano, flute and strings are quite different in this area no?

I mean all of them.

Piano, flute, strings, etc, Whatever they are, a staccato note should be played shorter than a legato one, isn't it?

Maybe I am wrong, because I have only learned to play piano. Hand should be lifted between two staccato notes. The same to other instruments. One person may take a breath between two staccato notes when playing the flute, or lift the bow and move the fingers when play a violin.

However I believe that this makes a natural effect.

Staccato is a completely variable articulation. If you are playing an instrument that is capable of sustaining a note it can be anything from a couple of ticks to nearly the whole of the original note.

Officially, or so I was taught, staccato should be performed by playing 50% of the note length: see this

90% seems way too long to me.

Maybe I used the wrong words. I don't know how to say that kind of notes in English.
What I mean is not the notes with a dot over its head. I meant the notes without a slur.

Hmm so you mean non legato?

It is possible to control the length of time a note is held by means of the note off offset in note properties.

I don't see how that could be automated other than by setting this - phrasing is a far more nebulous thing, governed very often by how the performer is feeling on that particular day.

Would be good if it could be though :)

Incidentally I'm not sure how to describe these notes in words either, and English is my native language :)

Staccato is indeed different from a "regular" note - that, a note not marked stacxato or legato. Staccato is normally a lot shorter than 90%. But for most people on most instruments, there is no real difference in length between legato and "regular"notes. Especially on wind instruments,the difference is often just a matter of how you use the tongue. On piano it's Morgan aiitutude than anything specific, but for many of us, Keaton's the norm, meaning notes get as close to 100%of their full value as is practical. When we see explicit legato marks, we may or may not actually *overlap* the notes. But this is all ery sibjective - it varies by the instrument as well the specific players.

Still, that doesn't mean it wouldn't be nice to have a dialog where you could control this more directly. Perhaps part of staff properties - so you could make it diffeent for each instrument - as well as global defaults. Righ tnow, I gather there are settings for this in instruments.xml?

You can already select individual notes and adjust the on/off times and you can select a group of nots or all notes and apply those settings. So when you find a setting that suits a particular instrument or piece (to your ear, at east) just remember it for next time and apply it so that you can hear your music the way that you want to.

Currently and AFAIU, this percentage is known as gateTime in MuseScore code.

  • Regular notes gate time : 100% and it's hardcoded
  • Articulations : For each articulations, we can define a gateTime in instruments.xml. The global ones, at the top of the file, applied to all instruments. Each articulation can be redefined in each instrument
  • Legato = slur = nothing particular is done

That much does seems sound, although before one can worry about that, slurs need to playback at all. I'm specifically talking about wind instruments here, where the difference is much more than just a subtle difference in length.

If you disagree with the original opinion, I mean disagree with the 90% way, just read the following stuff and you may change your opinion.

What I want to say is,
The Overture software does this in the 90% way (maybe around 90%, because I have never exactly measure it), for the notes described below:
1. nothing is over its head (like a dot or a ten. mark);
2. and there is no slur (or non-legato, whatever) or it is the end note of a slur.
I believe that some other well-designed MIDI software behaves in this way.

There are surely reasonable reasons to do this (maybe it is not suitable for some instruments, but fine for most).
1. Are there really no differences between non-legato and legato?
2. If there are differences, is the slur mark represents only for human?
3. If the answer to the one above is `no', so why don't the software produce a more realistic sound?
4. Every instrument have a slight break between two phrases, right? Not only piano, strings, flute (can you play the flute without breathing?), but also voices like choir.

Just consider the lines I wrote above.
If you do not think that some instruments require this, just make a `exception list'. But I believe that 90% should be the default.
And I will put a screenshot of Overture's graph analyze window which that kind of notes have the length of about 90% (or even shorter, but actually I think that 90% is fine and acceptable for most people).

The aim is that, software synthesizer should try its best to produce natural sounds, as real as they are played by human with real instruments.

Sorry, but I still I think 90% is way too short for default playback. I don,t care what Overture does; I care how real musicians play real music. The distinction that beginning piano students typically learn about betwen legato and ordinary playing is not one that most professional musicians I know would agree with - legato is the norm, not the exception. When we see legato marking, we might overlap the notes a little, but we don't deliberately leave space unless we see staccato or other markings.

For wind instruments, there is a difference between ordinary and legato playing, but the difference is not in note length at all - notes are always 100% length except when breathing. The difference is in the form of attack. Similar story for strings. But for piano, no professional musician I know woild deliberate leave space between notes just because they weren't marked legato. And in any event. Evem if one were to wish to play that way, it,s not the rule and every other instru,ent in the world the exception. It's the other way around - no instrument normally invites the possibility of leaving space *except* piano (well, and other keyboard instruments, of course). But I don,t think you"ll find a reputation professional pianist would say you should leave 10% space beween notes.

In any event, if you really want this behavior, you could create a plugin that shortened all actual durations. And if someone wanted to tweak MuseScore to allow someone so inclined to override the default length by editing instruments.xml that sounds like a nice enhancement as well. But no, I most definitively do not support changing the default to be something as unnatural as 90%.

Perhaps your are right.

However I wonder if there is a way to hear the effect of 10% reduced notes, because I do not even know if it is suitable for me.
I searched over my MuseScore package (I use Linux) and find nothing called instruments.xml
Please tell me where I can find my instruments.xml and how to edit it.

I think at this point, you'd be better off asking in the Support forum, since there are going to be a few different ways of going about this, and that discussion doesn't need to be part of this particular request. But note, I said that editing instruments as a way of shortening durations would be an interesting enhancement; it is not currently possible to change the defaults. That's why I suggested the possibility of creating a plugin. But again, this type of discussion is better suited to the forums.

My first instrument is Pipe Organ, and can definitely assert that 90% is way too short for normal playing.

To take an average hymn, the norm is legato - ie no gaps between the notes - in fact sometimes you even overlap them slightly (it is possible to produce a legato so glasslike that listeners find it uncomfortable). At the ends of lines and for emphasis, however, air is introduced between the notes to indicate to singers where to breathe. This reduction of the length of the note is completely variable, depending upon the emphasis required - notably if you are slowing down at the end of the hymn more air is introduced between the notes to emphasise the fact that the pulse is slowing.

In piano playing the 90% rule would work ok - but then pianists know very little about the proper articulation of notes as it is a percussion instrument, and it is done for them. Put a pianist on the organ with no prior training and the result is pretty comic :)

If you want to experiment with this - the parameter to play around with is the chord articulation in the Nightlies - note off offset seems to be currently disabled, although that will work in 1.2

It would be great if we could have a plugin which would apply a default to a selected range of notes, so you could select an entire phrase and have Musescore insert a breath at the end of it :)

Two points to clear:
1. In MIDI files a lot of the people sequencing music used the trick of shortening note to 90% or even less to have a distinction between slurred and "tongued" (for wind players). Also for piano music I saw this trick when opening a MIDI file in MuseScore with default settings of shortest note(1/64): a lot of short rests appear making the score unreadable. It helps to make the music sound more snappy rather than slouchy with the default decay.
2. The plugin suggestions do disregard that it is of very limited capability only. The note length is *not* part of the note properties exhibited. And with a plugin you cannot change ties (no setter a getter only) or create slurs. I look foreward to the plugins arising proving my statement to be wrong.

And I tried to export a MIDI from Overture software and import it into MuseScore.
I calculated the length of the notes:
Notes in a slur has a length of 100%.
Notes at the end of a slur has a length of 93.75%.
Notes not belong to a slur has a length of 93.75% too.

In the attachment is a demo, stored in Overture format (which can be opened in MuseScore), and a MIDI file exported from Overture.
You can open the Overture file, and see what the score is originally like.
And import the MIDI into MuseScore (and set the shortest note to the shortest one--1/128), then see what the playback of the originally score is like.

Right now, instruments.xml explicitly sets gateTime for normal articulation for 70%, which is WAY WAY WAY too short. Not sure if this was done in response to this issue, but I'd love to back this out. I could experiment with 90%, 95%, etc, but definitely not 70%!

BTW, my comment above relates to piano specifically. Other instruments continue to use 100 by default. Not sure that's really an appropriate distinction to make - piano isn't really unique here.

Playing around with this a bit, my proposal is we set the global default to 95%, and let specific instruments override this if appropriate. It's a subtle difference from the 100% "legato" setting, but in my experience, that's quite accurate. For most instruments / most styles, the default mode of playing really is quite legato, with an explicit legato marking indicating a softening of attack more than a different in length. For piano, legato might mean a small amount of overlap even.

I welcome feedback on this. I'd love if others were to experiment here so we can come to some sort of consensus.

BTW, I am testing this by saving a score as MSCX and editing the articulation setting there - it's much more direct than trying to get changes to instruments.xml to propagate into an existing score.

I've just been checking Marc, and the 70% Gate Time setting for piano goes back to version 1.2 at least (the oldest version I have access to).

It does seem very low!

However, a global default of 95% also seems wrong.

That would mean no legato was possible without a specific marking, which is definitely not the normal way of playing.

It would make more sense to leave the Global setting at 100% and configure the exceptions at 95%

Hmm, you're right that instruments.xml *says* piano is 70% even in 1.X, but is it possible this was being ignored until recently? Create a score from scratch, enter some quarter notes and eighth notes into it, and listen to the results. At least on my system, it's ncie and legato on 1.3, and that same score loaded into 2.0 is also nice and legato, but recreate the same score from scratch in 2.0 and it's basically all staccato. So it seems something changed where that 70% is finally being honored.

As for my suggestion of 95% as a global default, I was trying to find a compromise that allowed normal playing to be "fairly" legato but for an actual legato marking to make it *more* legato. Have you tried it? Maybe something else in the 96-99 range would be better, but 95% is still pretty smooth for most instruments. Really, I'm happy to leave it at 100% and accept that an explicit legato marking will continue have no effect whatsoever, but I'm open to having normal be *slightly* less legato than 100 as long as it still sounds natural. If you try it, I think you may find 95% isn't so bad, but if people find 97 or 98 a better compromise, that's cool too.

Here are two very simple examples. "Piano Articulation" was created in 1.3. Load it into either 1.3 or into 2.0 and it sounds normal. "Piano Articulation 2" was created in 2.0. Load it into 2.0 and it sounds staccato.

Looking at the contents of the MSCX, I see no gateTime in the PIano instrument definition for the 1.3 version, showing that indeed it appears to have been ignored in the instruments.xml. But the gateTime of 70 is present in the instrument definition for the 2.0 version.

So I'd say it appears a "fix" must have been made to the instrument list handling recently such that for the first time, instrument-specific overrides to gateTime are now being honored. And this fix has exposed a long-standing bug in the definition of the piano instruments that until now has gone unnoticed.

Attachment Size
Piano_Articulation.mscx 6.9 KB
Piano Articulation 2.mscx 6.26 KB

It is maybe like 'human playing' of finale.
It should render sound as we (musicians) play them....
but when exported to midi , it may not affect it, as when reimported the values are "unreadable" (musically speaking)...........
What do you think of it ?

Overall, I think global defaults for normal, legato, and staccato playback are fine, with just a few instruments needing overrides. Since no one complained about 50% staccato in 1.3, I'm inclined to leave that globally. 100% was the old default for "normal", and while few complained, this issue is one example. I think it needs to stay 100% for wind instruments since we don't have true slur playback - see #9345: Playback of slurred notes. With 100% as default for normal, playback will still have an attack (like tonguing) but no gap (like slurring) and serves as a best-we-can-do-for-now compromise, I think, until we implement true slur playback.

I'm still unconvinced there is any real value in reducing the default normal below 100% for piano. See #24246: Playback of piano notes is non-legato. If we reduce it at all, it shouldn't be below 95%, but it actually comes off a little uneven, since slurs do trigger legato, but on in the voice the slur applies to, whereas a real pianist would interpret a slur as applying to all voices.

So at this point, I actually propose 100% default normal as well as legato, 50% default staccato - for all instruments. In other words, exactly as it was for 1.3 as I understand it. If we need a threshold for short staccato notes, so be it, but maybe that's the composer's problem if he notates something staccato that is already so short?

Status (old) active patch (code needs review)

In 88d6be3666, all the piano instruments have a gateTime of 95%. Slurs will make 100% for all the notes under the slurs except the last one, for all voices in the same staff. Please test.

It's clever. My initial thought is that 95% for piano is still too short for me, but I need to play with it more

I'm not sure there will be a good solution here. The problem is that *if* one uses slurs consistently in a piano score, then one may want to hear a difference between slurred and non-slurred playback. But many scores won't use slurs at all, or maybe only a few at the beginning to convey to the basic intent, and then leave it to the performer to decide. So it really isn't correct to assume the absence of a slur indicates playback should be anything but completely legato. This assumption is valid only if the arranger is really consistent about his use of slurs.

So ultimately, I think we need there to be a user-accessible way to override whatever the default is for "normal" articulation. I guess we could go with something crazy like "use 100% as default unkess there are slurs present anywhere in the piece, in which case use 95% for non-slurred notes". But, "ick". Someday maybe we could add a dialog where user can set a few playback-related parameters, and this could be one.

Meanwhile, so far I lean toward going back to 100% by default in instruments.xml, but keep the code for slurs in place - that's the right algorithm, I think.

I support creating some GUI options to allow the user to decide how the playback should affect his/her score. IMO this options should affect the single score and not be general options (not affecting all MuseScore).
Even better, they should be measure/staff options: this will allow placing a textual marking like "legato" or "non legato" and then ordering the program how to play those particular measures the way you like.

In any case, I agree that there should be no hasty/ventured default setting. The user should decide, case by case, how to playback.

And in general, there should be a lot more options for playback in this regard. Not only about slurs, but also about dynamics and expressions, in Sibelius' fashion. The user could select a measure and order the program to play it "espressivissimo", so the notes velocities will change creating a simple crescendo and decrescendo for every group of notes inside a slur.

I would also like to point out that legato is not only about the duration of the single notes but about the dynamic too. Playing non-legato means playing a sequence of notes with velocities like 60 54 63 52 68, etc: basically random, although in a given range. Playing legato means playing the same sequence with velocities like 52 54 60 62 60 55 53 51 49. Basically creating a little volume curve. I'm describing phrasing with MIDI: this has its own limitations and it's just to give an idea of course. I just want to highlight that duration of the note it's not the only thing to consider, especially if we want to obtain an espressivo playback.

In reply to by Marc Sabatella

But what if the piano plays like Hungarian Rhapsody No.2 Friska? The pianist's left hand jumps like a frog but when they play slurs, its smooth. There is some difference between them. I agree with the title