Problems with ottava in cross-staff voices

• Aug 29, 2018 - 06:14

In a two-staff keyboard layout, if there are two voices in the upper staff, one of them a cross-staff voice originally belonging to the lower staff, the ottava sign applied to the first staff doesn't hold for the cross-staff voice. I'm not sure if this is a bug or it is intended, but it is not the expected behavior. The ottava should affect every single note present on the corresponding staff.
By the way, a similar problem does occur with accidentals.


In reply to by Jojo-Schmitz

Then how would one proceed if some notes from both voices require the ottava?

In the example, the first version emphasizes the fact that there is a middle voice that jumps from lower to upper staff. The ottava that this middle voice requires lies over the lower staff, so the third voice is in the incorrect octave. It would be possible to write this voice an octave lower, such as in the second version. It sounds as intended but it looks awkward. The third version looks better and sounds good, but the middle voice is no longer a single voice but parts of a voice of the upper staff and a voice of the lower staff.

In all cases there is still the problem that the ottava for the lower staff is not above the notes which it affects. It could be just moved up, but the upper staff would be crowded with ottavas, which is confusing.

Attachment Size
Test_cross-staff_ottava.mscz 12.11 KB

I tend to agree with fmiyara that notes that appear on a staff should be affected by other modifications on that staff like ottavas and accidentals. Perhaps a pianist could tell what he expects, but I would always expect consistency on a staff.

The usual recommendation is to simply not do that, since it is going to be ambiguous. It is better to write such passages at pitch, or if necessary, add a third staff.

In reply to by Marc Sabatella

The main reason for using cross-staff passages is
1) To keep trace of a single line or voice even if some notes are played by a different hand,
2) To avoid the need of using two voices (one on each staff) for what is a single line, and populating each of them with too many rests.
That said, sometimes you cannot "simply not do that". I would reserve the heroic solution of three staves for cases with too intricate a texture. In most instances it is not the case.
As an example, see this excerpt from Prokofiev's sixth piano sonata:

Clearly the left hand is at pitch, while the right hand is one octave high (you can listen to this played by Richter at at time 5:11). However, the whole line belongs either to the upper or to the lower staff, so if the theory expressed above were correct, the ottava should affect it all, so for instance, the B on the left hand after the ff should be played an octave higher, but it isn't.
I insist that the correct way to understand an ottava is that it affects only the staff on which it is and all of what is in its range, regardless of the staff where the line belongs.
So it is either a programming bug or an error about how to interpret the reach of an ottava sign. It should be corrected.

In reply to by fmiyara

To be clear, I am not saying to avoid cross staff notation. Just to avoid ottavas in that case, since the result is ambiguous and will likely confuse human musicians as well. But if there is a definitive source you can cite on the "correct" interpretation, we can certainly consider changing the default behavior.

In reply to by Marc Sabatella

To avoid ottavas is not a solution, they exist for some reason. In the Prokofiev example it would be necessary 5 to 7 ledger lines. Appart from the reading difficulty, it would take too much space. Besides, this is not about "good editing practice" but improving MuseScore.
As to a definitive source, the only book I have is Read's "Music Notation", and he doesn't mention this case. But I have provided one example. There are many others. An interesting one is that of Scriabin's tenth Sonata (,Op.70(Scriabin,_Aleksandr)), where you can see that even using three staves (he is known for using freely the three-staff layout) the essential problem persists. In the section headed "Puissant, radieux" the middle and upper staves contain cross-staff material (several written chord trills or tremolos using both hands). The upper staff has an ottava, the middle one does not, so the ottava does not apply to the whole line but to the upper staff only. Another example is the interludium between Fuga octava and Fuga nona of Hindemith's Ludus Tonalis. See 35:33 of, also by Richter.
Once the theory is dismantled by several real-life examples by authoritative contemporary composers and pianists, and having presented a breach of logic (impossibility to know, from printed music, to which staff originally belongs a passage containing cross-staff material), I think the only possible outcome is to try to fix the problem.
The expected behavior is that the ottava sign on a staff affects everything on that staff, and only what is in that staff, regardless of what voice it belongs to or if it originated in the other staff.

In reply to by fmiyara

I'm not saying avoiding ottavas is the "solution" - I am saying it is excellent advice. Because even though you personally might happen to feel that an ottava should apply to all notes on a staff even if some come from other staves, there is no guarantee the musician reading the score will share that same interpretation. That's what I mean by "ambiguous". It's a rare enough care that even if some particular author says it should be interpreted that way, many musicians won't be aware of that author's writing.

None of which is to say MuseScore's playback shouldn't respect whatever the officially correct interpretation is determined to be. But Gould doesn't given an explicit example of this (just stresses the need to avoid ambiguity), and apparently Read doesn't either. It would be nice to find someone who states a rule explicitly.

  • One thing Gould does explicitly say is "When octave transposition occurs only in one hand of a common beam, individual octaves signs should enclose only the relevant notes, so that the notes of the other hand are not mistakenly included in the octave transposition. When it is impractical to use separate octave signs, add loco ('at pitch') to the stave that is notated at sounding pitch".

In reply to by Marc Sabatella

Thank you for providing further opinions, particularly from Elain Gould's important book. However, there are many examples of celebrated composers that write their music as I "feel it should be", and piano players don't find any ambiguity. I think the practice of first-rank composers should be taken more as a guidance than the opinion of a music editor. Note I'm not discrediting Elain Gould, indeed she may be right about the ambiguity that some pianists may feel out of their lack of knowledge. But she is not telling how it should be interpreted. She is in denial as to the fact that this kind of writing exists (at least from your citation).

By the way, the use of the word "loco" is in the same category as courtesy accidentals. They make reading easier, but they are not mandatory. That said, I would certainly use it in Gould's example, but to make it easier the life of the musician.

Logic and corpus suggest that, even if it might be ambiguous, the behavior should be the opposite from what Musescore does. The same passage could be written in two different ways (two voices on the upper staff and moving some notes to the lower staff or one voice on each staff and moving the other notes to the upper staff, in both cases with an ottava at the upper staff), presenting exactly the same graphic result but sounding differently. This is not ambiguity, it is inconsistency.

A new example by Bartok: his Third Piano Concerto, 2nd movement. rehearsal mark 84

These findings are not just luck. Except an unsuccessful attempt from the Beethoven's sonatas set, it was easy for me to find examples from major contemporary composers because it is how, in the absense of some normative directions, they do proceed. To convince myself that I'm wrong I would need examples where first-rank pianists read differently a passage as the ones I cite (or similar).

In reply to by fmiyara

It's worth pointing out than unless you are consulting original handwritten manuscripts or "Urtext" editions, you really don't know how these composers notated it themselves. You've mostly only seen the editions created by unknown editors who I would not necessarily consider ahead of Gould. And even if you are looking at Urtext editions, composers are not necessarily expert engravers.

Anyhow, I don't question that there exist examples where this is the intended interpretation. I also highly suspect there are examples of the opposite. Gould isn't really taking a strong side on which is the correct interpretation, she is basically saying please don't do this because it is ambiguous, meaning it could reasonably be interpreted either way, and I absolutely agree with the single most respected authority on notated music in this century on this point. I have no problems with changing to a different arbitrary and subjective interpretation if the consensus is that this is more often preferable to the other equally valid arbitrary and subjective interpretation, but let's call a spade a spade here: this is not a case of clear-cut right and wrong.

Addendum: I also don't doubt that in the particular examples you cited, the context makes the intended interpretation clear enough that most would agree on it. Plus of course there are earlier recordings and an oral tradition before that to provide supplementary confirmation. So it really isn't surprising there might be widespread agreement in these cases, based on the specifics of the cases themselves. However, it would be a mistake to assume this means anyone who played that particular passage that particular way would agree there is an immutable law on such matters in all cases.

In reply to by Marc Sabatella

There are two points you have not even considered in your otherwise thoughtful answer:
1) I have provided several examples supporting my point of view but have not seen a single one on the contrary; you just suspect there are. However, those who have programmed this specific feature for some reason have chosen a different interpretation. It must be based either on some evidence or on their personal opinion. If there is such evidence, I would like to see it.
2) I have shown that there is a logical flaw in the way the current feature behaves: there is inconsistency in the way the same printed score is played by MuseScore according to the way the engraver choses to set up the voices. Since this software is widely used in music teaching, it could cause confusion in the students and hamper reaching sometime in the future a normative interpretation of this particular situation.

In reply to by fmiyara

Indeed, I don't have examples. But that's because it is the very lack of examples that supports Gould's case. She states the rule as, "don't do this, because it is ambiguous", and the very fact that the vast majority of editors follow her advice and don't use that ambiguous notation is evidence that she knows what she is talking about. The fact than a few editors have either not known the rule about avoiding that ambiguity or have chosen to ignore it in hopes the context clarifies the ambiguity in their particular cases doesn't disprove the rule. Really, the only way to prove her wrong would be to find an example of someone who is an equally respected expert in 21st century engraving explicitly saying this is not ambiguous and that there is One True Interpretation, and editors should feel to notate with the expectation that if you notate that way way, it will always be 100% clear.

I"m not understanding your second point about the inconsistency, however. If you'd like, feel free to explain that in more detail.

Anyhow, to me this is way way way overthinking things. Gould clearly says not to do this because it is ambiguous, and that remains excellent advice. Someday if enough demand builds we might change the default playback behavior when people chose to break that rule and create this ambiguous notation. But for now, people who chose to break that rule and create ambiguous notation can use a workaround , like say the "Fix to line" option, if it is important to them to have MuseScore interpret this ambiguous notation in the way they prefer. For the number of times this comes up, it's probably not worth the amount of time we've already spent on this...

BTW, as for why it was programmed like this, mostly likely it is for simplicity. The cross-staff notes "belong" to the other staff, so at the time we are processing playback, it is the ottavas on that other staff we have easy access to. It would require extra programming to go out of way to check ottavas on the current staff. It's not impossible, but would require special-casing. No doubt this wasn't added because it was assumed it wouldn't matter.

In reply to by Marc Sabatella

Please, again, note that I'm not talking about "good editing practices", with which I agree, but how MuseScore renders some writing. The fact is that even in cases that aren't ambiguous at all, the result may be paradoxical or inconsistent. See the following example, the one that motivated my first post:


As you can see, both versions look the same, no one is ambiguous, i.e., no pianist would have the slightest doubt how to interpret this. Yet, they sound different. Why? Because in the first (the correct one) I have forced the intermediate voice with the arpeggio to be on the upper staff, moving the three lower notes to the left hand staff so that the ottava affects both voices on the same staff, while in the second one I proceeded the other way round (as I would prefer, since the original voice belongs to the left hand). The result is that what is written as a unison (top note) plays incorrectly as an octave, since the ottava applies only to the voice "genuinely" belonging to the upper staff.

As to the programming difficulty, of course any enhancement or new feature requires some extra programming, otherwise the software would freeze and stop being maintained. Needless to say, not being one of the programmers I cannot speak with any authority, but looking at the .MSCX file I see that cross-staff notes bear an attribute 1 or -1 according to the direction of the cross-staff operation. This could be used to filter which notes receive the ottava. If a note on staff n has the the move attribute present, it would be necessary to check whether at staff n + move there is currently an active ottava.

In reply to by fmiyara

I realize you aren't talking about "good editing practices", but to me that is an inseparable part of the topic. For notations where there is a clear-cut universally-agreed-upon "correct" interpretation that is supported by the experts in music notation practice, then of course MuseScore should strive to get it right, and those cases should rightly get higher priority. It's the cases where the experts say "please don't do this, it's ambiguous, musicians might well interpret it either way" where it is less obvious that MuseScore should take resources away from other features just to change the way we currently deal with it.

Once again, I'm not standing in anyone's way who feels that changing this is the most valuable use of their programming time. I'm simply observing why I personally don't think it is. I'd rather see focus on notation over playback, and to the extent we focus on playback, on the places where a universally-accepted notation is not interpreted in the universally-agreed-upon way. If others feel differently, they are free to allocate their programming time accordingly.

In reply to by Marc Sabatella

You keep stressing on the "don't do this please" directive in cases of ambiguity. I have acknowledged that ambiguity should be avoided (when it is really ambiguous) making life easier for musicians. But I have also provided real-life examples where there is no ambiguity at all, yet the way MuseScore renders them depends on non-notational matters. My example also satisfies the "you can do this" criterion given by Elaine Gould (see example (a) from your citation) so there is an authority telling there is no ambiguity. Just in case you haven't opened the mscz, this is one of such examples:


This example alone should convince you that there is a problem.

In reply to by fmiyara

I am not stressing "don't do this please"; I am simply using the fact that this is what the most recognized expert says as indication that this is indeed something that is known to be ambiguous and thus not something where there is a clear and obvious right and wrong answer and clear and obvious reason to take programming resources away from other features just to make this work differently. Again, I'm not opposed to it, I just think many other things would be higher priority, precisely because this is acknowledged to be bad practice that can lead to ambiguous interpretation. Once more, if someone who has programming expertise and extra time on their hands but who can't think of any actual recommend notations to spend their time on wants to take your advice and spend their time working on changing this, I certainly am not opposed to it.

I've said the same thing many times now, not sure how yo make this more clear.

In reply to by Marc Sabatella

This is a follow-up of a rather old post.
I've been able to contact Elaine Gould through the publishing company where she works, Faber Music, and asked for her opinion. It took some time but I've finally get her answer. I've asked permission to share it, but I have no reply yet. However I can anticipate that her opinion regarding the example I posted on Sep 1, 2018 - 17:22 is the same as mine: when non-ambiguous ottava is located over a staff it affects everything on that staff, regardless of the underlying structure, i.e., whether part of the notes originate in the same staff or in the opposite one using the cross-staff tool.
To be honest, she also insists in avoiding ambiguity and provides good additional insight supporting her recommendation and analyzes convincingly the examples I've provided her. But the case I'm talking about is not ambiguous, yet MuseScore interprets it the wrong way.
I have been also told by a friend who uses Sibelius that this weird behavior undermines also Sibelius, so it would be a good idea to fix it and so have a better rendering than Sibelius :)

In reply to by fmiyara

New follow-up
I've been authorized by Elaine Gould to share her insights on this issue, so I copy the relevant parts of her reply. The comments by Elaine Gould have been separated for clarity as if it were a dialog, but they are originally in-line comments in other color:
"Dear Federico,
Thank you for your enquiry. Answers below. I recommend using a precautionary (loco) at all times for the stave not affected by the 8ve sign, even though this hasn’t necessarily been used in the past.
Please do not hesitate to contact me if you have any other queries.
Best wishes,
Elaine Gould
FMIYARA: Dear Elaine Gould, I'm writing you through Faber Music trying to elucidate a particular notation issue which is not completely dealt with in your book. It refers to how should be interpreted a piano passage containing a cross-staff beamed voice along with a non-cross-staff voice in the presence of an ottava transposing symbol, such as the following one:

(F. Miyara, Sonata No. 4) [NOTE: The example originating the post and presented in my post Sep 1, 2018 - 17:22]

The issue arised when using score editing software such as MuseScore (a free software) or Sibelius. These applications, which also allow to get playback of the music, present an inconsistent behaviour. If the intermediate voice (the arpeggio) is originally a voice belonging to the upper staff, the ottava will be applied to both voices, so the F flat at the top will be played as a high-pitched unison. If the voice originally belongs to the lower (left hand) staff and the top notes have been moved to the upper staff, only the top voice will be affected by the ottava, so the upper notes of the arpeggio will sound loco and so the unison will sound as an octave. I think this is incorrect.
FMIYARA: I believe the ottava sign should hold for all notes belonging to the corresponding staff regardless of their structural origin.
FMIYARA: My first question is if you can confirm this.
FMIYARA: A related situation is the one you suggest as ambiguous and should be avoided:

[NOTE: The example cited by Marc Sabatella from Elaine Gould's book, see Aug 30, 2018 - 15:46]

ELAINE: WITHOUT ‘(loco)’ for the lower stave I think it is ambiguous.
FMIYARA: However, there are several examples in music of the 20th century, some of them from very famous and known works, where similar examples are used.

[NOTE: Some examples that I've provided earlier in the post]

ELAINE: YES INDEED, but I recommend what I think is best practice. I think all the examples you quote are unclear – if we were unfamiliar with the music of these composers we wouldn’t necessarily know which octave was meant. The Bartok is very ambiguous. Why does the octave sign appear to start for the LH notes?! This probably indicates the setter wasn’t a musician. The Skryabin octave sign is also placed too early. Luckily it’s obvious here from the musical context that the LH is loco!
FMIYARA: These are examples where musicians consistently interpret the passages using the criterion I have exposed above i.e., the ottava sign affects everything on the staff to which it is applied and it doesn't affect the cross-staff notes under the same beam. I've reviewed many youtube videos where, when in doubt as to what is heard because of dense texture or less-than-perfect live recordindgs, the hand position confirms the way they interpret the score.
ELAINE: If the score were clearer you wouldn’t need to check.
FMIYARA: I acknowledge that some pianists will be in doubt and will have to resort either to earlier renderings or context or analysis, but the result is always the same, so I think there should be an "official" or "normative" way, and palliatives such as the word "loco" would be in place at the same level as precaution accidentals or courtesy clefs.
ELAINE: From what you’ve cited I’d say that loco was more important than just a precaution.

In reply to by fmiyara

Thanks for the followup! I'd say it does clarify that there is a technically "correct" interpretation, but also that it is not advisable to rely on this because it is unclear. It doesn't change anything about my personal assessment of the priority of this, but it certainly removes any argument to the effect of "what if it's the wrong thing to do".

In reply to by Marc Sabatella

You seem to be talking as if both clauses applied to the same situation and I'm afraid they do not.
What Elaine Gould actually says is that when there is no ambiguity, the technically correct interpretation is that the ottava applies to everything on the staff to which it belongs--and there is nothing unclear about this. What is unclear is the way MuseScore plays it.
And, on the other hand, when there is ambiguity there isn´t a technically correct interpretation and her suggestion is to use the word "loco" attached to the other staff.
It is the first case that worries me the most, because there are many teachers around the world using MuseScore to teach Music Theory and Music Notation and the students could be confused by an inconsistent behavior.
I agree it probably is not top priority, but it is certainly a bug and, as per the accepted definition, a "major" one (a common feature --the ottava-- incorrectly functioning).

In reply to by fmiyara

Somehow I am not understanding the distinction you are making between which cases are unclear and which are not. Basically, I take the comments as saying that while technically ottavas apply to everything on the staff, any time there is cross-staff notation there is potential for confusion. In some specific cases the context might help human musicians disambiguate, other times it will be less clear.

In any case, even if we could state a clear rule for which cases should be considered clear and which should not, it still doesn't change anything. I agree that now that we have a definitive answer as to what is technically correct, we can classify this as a bug. But as the person who helped develop the "major" classification, I can assure you this is not what it is intended for. MuseScore is a notation program, and the ottava feature is in fine basic working order with respect to notation. Some particular detail of the playback is not quite right, but that's not nearly the same as saying the feature is fundamentally broken to the point of being unsuable for the main use case it is intended for. Even if the playback of ottavas didn't work at all this would be no worse than half the other markings on the palette. As it is, playback works perfectly for probably 99% of cases. Only this one small detail is off, and it is off only in cases that acknowledged to be potentially unclear depending on more specifics of the context. This is, if anything, minor according to our classification system. But if someone were to submit this as a bug report and call it "normal", I won't quibble that.

In reply to by Marc Sabatella

OK. This is a case where no ambiguity exists, it is clear for 100 % of the musicians (Elaine Gould included), yet MuseScore plays it the wrong way if the internal arpeggio is originally part of the lower staff and the top notes are cross-staff:
One of the reasons why MuseScore is so widespread compared with LilyPond, for instance, is that it can actually play the music (another is being WYSIWYG, of course). Nowadays, a notation program is not just one that replaces the tedious engraving craft at a specialized facility but one that also allows listening to the music. That's why all comercial software adopts this policy: it's more competitive.

In reply to by Marc Sabatella

Here is another example, Bartok's Third Piano Concerto, coda to the finale:


I edited three versions with MuseScore. For the first one I wrote the expected notes on the lower staff, then moved to the upper staff the notes that in the original score go there, then lowered them an octave and applied the ottava sign. As the ottava is automatically placed over the content of the measure, which in this case are measure rests, I had to move its ends to get the same look as in the original.
In the second version, i did the same but starting on the upper staff and moving the other notes down to the lower staff.
In the third one, I kept the six ledger lines to avoid using ottava.
As can be checked, versions 1 and 2 look the same but yield different musical results (this an unacceptable inconsistency). None is the intended one. Version 3 sounds as it should (and as one can see pianists perform in youtube videos), but it doesn't present the original layout.

In reply to by Marc Sabatella

The case of Ravel's Concerto in G is interesting (it can be downloaded from IMSLP). It has several passages combining cross-staff notes and ottava; for instance, first movement, rehearsal marks 5 and 6. Here he (or his editor) solves the ambiguity by using an angled line, instead of a straight line, with segments surrounding the notes to which the ottava applies:


This would be an interesting addition to MuseScore, but it would take a fair amount of thought to design an algorithm to make it reponsive, i.e., that it sounds what is intended.

In reply to by Jojo-Schmitz

Giving ottava lines the same options as hairpins, it wouldn't help. With all of these cross staff notation examples, the ottava only applies to the notes that appear on the top staff. In many of the examples there are notes from both staves on the top staff and more notes, not affected by the ottava, on the bottom staff. If this were not the case the workaround would be quite easy.

In reply to by Jojo-Schmitz

In my recent reply to Marc I suggested a way to deal with the problem.

However, as a suggestion inspired by your comment, it might be useful to introduce a new type of range setting: voice-wise. This could provide a solution to a situation where on the same staff one has two different voices, each with different hairpins.

In reply to by fmiyara

Making some dialog that allows you to choose the voices in a grand staff that are affected by an ottava or for that matter any hairpin, would be a good way to make this work.

From a notation point of view, MuseScore does perfectly. Interpretations of symbols such as the ottavas in cross staff notation are subject to some interpretation, bu I fmiyara is correct with what is expected most of the time. There seems to be some emphasis being added to playback in the future of MuseScore and allowing for the different interpretations of such symbols would be useful.

I seem to have a smilar but opposite opposite problem - I have cross staff passages originating from the top staff, but extending to the bottom staff. There is an ottava on the top staff which should ONLY apply to those notes on the staff, but it extends down to the notes on the bottom staff during play back. Please see example.

Attachment Size
Octave.mscz 10.59 KB

In reply to by Bryan Saylor

This is a case which, according to Elain Gould, would be considered ambiguous and would require the word "loco" (at the written pitch) written on the lower staff. However, there is either no way to honor such indication (one possibility would be to take it as a staff or text property similar to "pizz." on string instruments).

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