How to correct/repair my note sheet ?

• Nov 3, 2020 - 18:22

Hi,

I use a scan app to scan a note sheet. But seems there is some error by this app and the scanned note are not 100% correct. So I try to use your software to "repair" it. You can see this in my attached file.

Regards

Attachment Size
question.png 34.12 KB

Comments

To add line breaks, hit Enter.

To insert notes, in note input mode use Ctrl+Shift+letter. But, I suspect your problem here is deeper than that, those are probably meant to be triplets, and I think your scanning app missed that. There is no easy way to correct that error aside from deleting them - really the whole measure, since it is also now too long - and re-entering them yourself.

In reply to by Marc Sabatella

Thanks for the prompt answer!
1.
The line break is so easy!

2.
You are right. It should be a triplets. I've written to the developer of the app. He admited their app is not 100% perfect yet.

Complete to delete it terrible. :-) Can I just delete a section and insert it?

Besides the note sheet is in xml format. Can one edit it? (the problem is where can I find its location in the xml file)

Regards

In reply to by Jojo-Schmitz

attached is the original one I scanned. But the note scan app can not implement such "complex" music notes.

Besides, I am not sure if I can use the musescore to edit such "complex" notes? Even it can, it will take me a lot of time. I use this note sheet for my daughter. I think it is enough at the moment.

Attachment Size
page8.png 79.3 KB

In reply to by thomas2004ch

There's nothing any scan App can do with things that are just omitted and 'assumed', like here the triplet brackets in 2nd and 3rd measure of 2nd staff. Software is really bad at doing guess work.
1st measure in 2nd staff might work, but that triplet notation is somewhat unusual, using something thal looks like slurs, so I'm not surprised if a scan app gets that wrong too.
Howver: these few measures are just not worth it to get scanned, you'd have entered them into MuseScore in 5-10 minutes, on a bad day

In reply to by Jojo-Schmitz

The slur type triplet brackets are archaic and cause confusion to humans. Besides fixing wrong notes I think deciding what to do about slurs on triplets is the most time consuming part of reviewing scores for the Lieder Corpus, which uses scores from ca. 1780 - ca.1920 and we rarely see square triplet brackets.

In reply to by thomas2004ch

The pdf conversion service that MuseScore uses
https://musescore.com/import
relies on the Audeveris software. You can download the software for yourself (to tweak its recognition parameters for greater accuracy, something not possible using the conversion service):
https://github.com/Audiveris/
or (simpler to understand):
https://audiveris.en.lo4d.com

Over the years, I have tried a few OMR - optical music reader - applications. There is a cost/benefit issue to consider.
Sometimes I would get much better results doing some "clean up" of the original image:
Here's your original scan attachment comparing A (original) to B (cleaned up):

scan_version.png
An OMR will give much better results for B right from the start as there is less "chatter" - e.g., text, hand written markings, lines, symbols; and there is more "meaning" - e.g., notes, staff lines, barlines.
The eye/brain combination of a human musician is much quicker at discerning the difference between "chatter" and "meaning" than is a machine. To a machine, every pixel of an image is neutral (in meaning or chatter) upon its first encounter. Because of this, OMR software typically has parameters that can be set so that on successive readings its accuracy improves.

For example, the "complex" notes you mention are triplets. Music scanning software should recognize a time signature and "knows" when a measure is over or under populated with notes/durations. In your case some may infer triplets are needed. Some will show a little + or - to mark those irregular measures. Issues like this can be handled within the OMR and the software can produce more accurate results after another iteration.

(MuseScore produces something similar:
https://musescore.org/en/handbook/measure-operations#duration
Look at your 'question2/png' attachment and you will see + marks at the end of the measures that should have triplets.)

As MuseScore is not an OMR, the idea is to have the OMR produce accurate output before it gets loaded into MuseScore (where the three-note non-triplets can be fixed, although now with greater difficulty).

So...
You can fiddle around with OMR, or learn to enter notes into MuseScore. The choice is yours.
Viewing your posted images, (they look like simple musical exercises) note entry into MuseScore would be easy enough, with a very short learning curve.

Regards.

P.S.
The image I used is a very simple example. Usually I'd get rid of things like lyrics, handwritten instructions, extraneous printed text, smudges, even sometimes deleting entire stave(s) or system(s). Then re-scanning into the OMR.
(Again, cost vs. benefit.... Yay! After 2 hours, I got the OMR to recognize that time signature change!)

P.P.S.
I've noticed that OMR software is always improving...
Nowadays with all the submissions to the MuseScore service, I wonder if some deep learning is occurring.

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