Is it possible to exterminate notes and rests?

• Sep 16, 2020 - 21:35

I know, this must sound weird, but it's still a serious question. ;-)

From time to time I work on scores that were converted from PDF to MuseScore. In the converted .mscz file I sometimes see measures with more notes and rests than the time of the measure allows. As an example, a "Bass clef" halfway the original PDF is sometimes recognized as a "Whole note" and added as another note with the other notes into the measure in the .mscz file. Yes, then it gets crowded there. (Excuse me, would you please take your elbow out of my eye? :-) )

I could correct this in an easy way if I can "exterminate" the notes or the rests that are too much. In the example that I gave I would completely remove the "Whole note" that was originally the "Bass clef" and the remaining notes and rests would then add up to the correct time of the measure.

Is this possible?

If not I'd like to toss in a feature request.

Thanks to everybody answering this question !


Comments

The problem with such a feature is that knowing what the original score looked like is required. Why? Because (almost) everywhere there is a spurious rest there is a note that is too long by that amount. Sometimes the note is on the same beat, sometimes not. That can only be determined by looking at the PDF. The PDF is imported using audiveris which is developed independent of MuseScore and it what actually needs to be fixed. Their site is http://audiveris.org/

In reply to by mike320

That's pretty much how I do it now.

I have the original PDF and the converted .mscz on the screen at the same time. Then I compare the two and I notice that, by example, a whole note somewhere in the .mscz file was originally a bass clef in the original PDF. I'd like to manually take out that whole note myself.

Right now I insert a new measure, copy over all the correct notes into the new measure, and then remove the "old" measure. That however, is more work, especially if it needs to be done repeatedly.

I'll try to find an example.

I agree that it's better if this is corrected at the source, e.g. with Audiveris, but I don't know how doable that is for them. I remember the first OCR applications with their "children's diseases". ;-)

In reply to by barencor

The OMR (Optical Music Recognition) program needs to be improved and MuseScore does not do OMR natively at all so it can't compare the music to the PDF. What you are asking for is impossible unless MuseScore ventures into native OMR which I don't think will happen any time soon.

In reply to by mike320

Here's an example of the original PDF. The purple arrow shows the Bass Clef that I mentioned:
PDFwithBassClef.png

Here's the same part when converted. The purple arrow now points at the note that I want to remove. (I just want to remove it myself) :
MsczWithWholeNote.png

I am not suggesting or asking that MuseScore compares the converted music to the PDF. I am asking if there's a way that I can remove that note myself. If I delete it now, it's replaced with a rest of the same duration and the measure is still corrupt.

Attachment Size
PDFwithBassClef.png 18.67 KB
MsczWithWholeNote.png 23.14 KB

In reply to by mike320

The measure that I point to (i.e. measure 4) has a "half note" plus two "eight notes" plus a "quart note" in the lower bar. That's a total of one. Where's the missing beat?

Correct, Audiveris doesn't handle it correctly. But I still need to do something. For the moment the only way to get it straight, is to correct it myself in MuseScore and that's what I have been doing.

The fastest way is to "take out" the "note" that's too much..

In reply to by barencor

Oops, my eyes played a trick on me. It's still not MuseScore's fault and there isn't much MuseScore can do to help the situation because things are added in strange places by Audiveris and MuseScore has no idea what should be in the measure and what shouldn't

In reply to by frfancha

I had tried that. It replaces the "whole note" with a rest (of the same duration).

What I get is a measure with a "half note" plus two "eight notes" plus a "quart note" plus a "whole rest", all in one measure. That's a duration of two, while each measure should have a duration of one.

Not easily. Think of the measure as a container. The music scanning software has counted the notes wrong and made a bigger container to fit them in. Deleting notes still leaves the container the same size and it must contain something so it contains rests.

Easier to insert a new measure (of normal length) and copy the good notes (and rests) over to it and then delete the bad measure.

In reply to by underquark

Actually, this would be an easy change. You are right that the measure (or container) size would also have to be changed, but that means changing a number in a record or a field that represents the measure size and that's easy. In SQL terms that would be something like this :

update [table name]
set measure-size = 1
where measure-id = 4

(Note, "4" is the measure-number from the example)

But if the people behind MuseScore don't want to do this, that's fine with me. I understand that sometimes there can be other priorities.

PS:
- Inserting a new measure and copying the correct notes and rests over and deleting the "bad" measure is what I did so far, but it's a tedious process. Thank you for the hint anyway.
- I do think that MuseScore becomes a much more valuable product, for many people, if users can correct converted scores easier. ;-)

In reply to by barencor

That's easy for this measure but in general these extra beats can be added anywhere in a measure and eliminating a spurious 8th rest on beat 2 might delete a legitimate 8th note but it may shorten a note that needs to be shortened. The only way to know is to look at the PDF.

In reply to by mike320

That's easy too.

If I "eliminate" a note or rest, then anything following that note or rest would move forward.

Here's an example of another "corrupt" measure :
A-CorruptMeasure.png

I would like to select the note or rest that shouldn't be there:
B-SelectCorruptRest.png

Then I would like to push ctrl-del (or another key combination) and the "corrupt" rest would be eliminated. The notes behind it would move forward and my measure is now correct:
C-Removed.png

This also means that eliminating an 8th rest never deletes another, legitimate note. If there's a legitimate note after the "corrupt" 8th rest, then the legitimate note would move forward, to the position where the "corrupt" rest used to be.

It compares to the way that I remove an entire measure right now. When removing an entire measure, the measures behind it now also move forward.

With 90% of the the corrupt measures that I saw, there was something too much in the measure. The process would be roughly the same for all of them. In maybe another 8%, tuplets weren't recognized as such. In those measures i saw three regular notes, where there should have been one tuplet with three notes. Then there was one measure that contained the notes of two measures combined. In another case the 3/4 sign at the start wasn't recognized. That gives funny results too. ;-)

Mmm ... maybe I should have started this thread with this information and this example?

Let me know if you see any other obstacles. I'll be happy to resolve those too.

Frfancha, thank you for your last comment. I'll check later. I hope you are right.

BTW, it's fine with me if this suggestion is declined. However, I think that this suggestion requires low effort to implement, while it has high value for those converting PDF's to MuseScore.

Sometimes it's good to have a short break and give a topic a thought. I hoped that I would bump into an useful example, which I actually did. Also, by coincidence I noticed it's already possible to shorten a measure, although not in the way I need.

frfancha wrote:
> Yes it is possible. Just delete it in insert mode, or even
> when not in insert mode select it and press ctrl-del

I tried it again. Please try it with the MuseScore part that I included. If I hit ctrl-del in MuseScore3, then that sometimes works, but usually I get the message "Please select the complete tuplet and retry the command". That also happens when I'm trying to delete a note or a rest that's not part of a tuplet. I suspect that the tuplets on the other staves are "in the way". (Please refer to the image below.) I assume that when you tried it with your score, it worked with all the rests that you eliminated. Maybe we were both right?

mike320 wrote:
> That's easy for this measure but in general these extra beats can be added anywhere in a
> measure and eliminating a spurious 8th rest on beat 2 might delete a legitimate 8th note

Now I understand what you tried to write. Am I right that you mean a "legitimate 8th note" on the same beat, but on a different stave? If I am able to "eliminate" a note or rest by pressing ctrl-del, then indeed, I do see changes on the other staves as well. You're right, these changes can render a correct part of the score incorrect. I'm confident however, that there's a solution. If someone wants all notes and rests on the same beat to be eliminated, then draw a rectangle around them and press ctrl-del. ;-) If I only select notes and rests on one stave and press ctrl-del, I would like the other staves to be unaffected.

If I would be able to eliminate the selected rest in the next image, then the ties that cross the bar would also remain intact. That would be super nice.
NotesWithTies.png

I also found that it's already possible to "shorten" a measure. If I right-click in the bar and choose "Measure Properties", I get this:
MeasureDuration.png

In field "Actual" I can change 17 to 16, but this will "chop off" the last part of the measure of all staves. In this example that's not what I need.

I appreciate all the feedback and comments. Not only does it let me know if I have a reasonable request or not, the feedback and comments also let me know if I wrote things sufficiently clear, plus it lets me know if I overlooked something. Thank you all !

If I would submit a feature request, I'd have to write my request different.

Attachment Size
CorruptMeasures.mscz 17.38 KB

In reply to by barencor

We don't need to try again and submiting a feature request doesn't make sense I'm afraid.
You have indeed discovered and (almost) understood the MuseScore behaviour.

So to be complete but not too long:

When a score is made of several staves (basic example being a piano score with its bass and trebble keys), all measures are "aligned" between all staves.
So if one measure somewhere happens to be 5/4 instead of 4/4 for whatever reason, that will be the case in all staves.
Suppose you decide to remove beat 3 of that 5/4 measure by ctrl del, beat 3 will be removed in all staves.

For triplets, yes MuseScore is completely unable to manage partial tuplets (which is normal if you think about it). Therefore MuseScore doesn't allow any operation that would create partial tuplet. Sometimes that limit is too strong: you can not copy part of a tuplet and paste it on another similar tuplet where it could actually fit. In your case though the blocking is legitimate as deleting part of a tuplet would indeed create an unallowed partial tuplet.

P.S. to not make the text too long I have just mentionned "a 5/4 measure" : the explanation is valid for a "true" 5/4 measure and also for a "actual duration" 5/4 measure written in 4/4 as pdf scanning has created (or as you can create yourself).

In reply to by frfancha

Thinking more about it, a suggestion could be introduced:
-allow different actual measure duration by stave, MuseScore playback behaving as if the "global measure" would have an actual duration = max [ actual duration by stave ]
That way it would be possible to allow ctrl-del to remove different part of the measure by stave ...
Once the extra part removed on all staves, the 'formula' actual duration = max [ actual duration by stave ] would give the correct result
Implementing that suggestion would lead to potentially strange scores though, not sure it will ever be considered, but who knows

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