Extra 8th Notes but Measure Timing OK?

• Oct 4, 2018 - 19:55

The following 2 measures are excerpted from Verdi's Te Deum, which I recognized from a scanned PDF using PhotoScore 8 and then imported as XML to MuseScore.

Can someone explain what is going on in these two measures?
The upper staff in Voice 1 appears to have an extra 8th note, yet Measure Properties shows the length of the measure is properly 4/4.
I've tried deleting the notes, exchanging the voices, etc. but the extra 8th notes persist.
What now?

-Tom

Attachment Size
Extra_8th_Note.mscz 9.92 KB

Comments

• Oct 4, 2018 - 21:13

Thanks, Shoichi.
I tried your method in the file from which these two measures were extracted, and it didn't work.
The extra 8th-rests persisted.
I wonder why it would work in this extracted measure?
Here is the entire file, look at measures 135 and 136 for these two errors, if you please.

Also, I cut out the Voice 1 part and pasted it into a 3rd measure. The extra rests went with it.
When you look at View/Measure Properties, you see no error.
But if you close the file and reopen it, the box shows up that the file is corrupted and the measure is properly identified.

My concerns are :
1) What has caused this sort of corruption?
2) What can be done to fix the Measure Properties box so it properly shows these errors?

Attachment Size
Verdi - Te Deum.mscz 103.32 KB
Extra_8th_NoteBis_0_B.mscz 10.03 KB

Reply

• Oct 4, 2018 - 21:39

There is a lot of "artificial intelligence" that needs to happen in order to make the extremely experimental step of construction an actual musical score given only a picture of it (the PDF). It's basically like "Siri" attempting to understand your spoken word, or handwriting recognition. Sometimes it works well, a whole lot of time it doesn't though. In the end, it's often more efficient just to enter the information in more traditional ways.

Reply

• Oct 5, 2018 - 04:45

Marc,
I am in awe of what programs like PhotoScore and Smartscore can accomplish.
However, inevitably their recognition process leaves errors.
That is not a surprise to me, and I have gotten used to dealing with most of them.
My concern here is not the xml file but how MuseScore fails to show the error in Measure Properties.
Does anyone have an explanation for that, and why even deleting the erroneous voice part and reentering it does not eliminate the error?
How can can I use MuseScore to fix such an error?

Reply

• Oct 5, 2018 - 05:44

you're gonna have to eliminate and rewrite::
Bar 59 Stave 10 incomplete. Expected: 4/4; Found: 9/8
Bar 61 Stave 8 incomplete. Expected: 4/4; Found: 9/8
Bar 61 Stave 10 incomplete. Expected: 4/4; Found: 9/8
Bar 104 Stave 10 incomplete. Expected: 4/4; Found: 214/192

Attachment Size
Verdi - Te DeumTer.mscz 102.12 KB

Reply

• Oct 5, 2018 - 08:51

Shoichi,
Did you fix some of the errors in the file I posted here?
In my MuseScore file, I get additional errors:
File "/Users/.../Verdi - Te Deum.mscz" corrupted.
Measure 53, staff 11, voice 2 too long. Expected: 4/4; Found: 6/4
Measure 59 Staff 10 incomplete. Expected: 4/4; Found: 9/8
Measure 61 Staff 8 incomplete. Expected: 4/4; Found: 9/8
Measure 61 Staff 10 incomplete. Expected: 4/4; Found: 9/8
Measure 104 Staff 10 incomplete. Expected: 4/4; Found: 27/24
Measure 133 Staff 10 incomplete. Expected: 4/4; Found: 21/16
Measure 135 Staff 10 incomplete. Expected: 4/4; Found: 18/16
Measure 136 Staff 10 incomplete. Expected: 4/4; Found: 19/16
Measure 174, staff 9, voice 2 too long. Expected: 4/4; Found: 23/16
Measure 216, staff 3, voice 2 too long. Expected: 4/4; Found: 7/4

If I click on "Ignore," my file opens up and appears OK except for rests as reported above.
I wonder why Measure Properties doesn't show any of these errors?
All the measures show as 4/4, from beginning to end of the piece.
-Tom

Reply

• Oct 5, 2018 - 09:43

@tomfeledy
I can't answer your two questions, but I can offer a comment. Frequently over the past five years I have used OMR (SharpEye and PhotoScore), generally with good results. But it seems inevitable that corruptions will occur, as Marc explained.

The question is: how best to fix the corruptions? After a lot of experimenting, I have settled on two methods:

  1. For a minor corruption problem in one stave only: copy whole-measure rests into the affected measures and the following measure, in a single copy-and-paste. Then re-enter the missing notes manually.
  2. For a major corruption involving the same measure in many staves: entirely delete (Ctrl+Del) the affected measure(s) from the score. Then insert the missing measure(s), and re-enter the missing notes manually.

I avoid copying any measures in the hope of saving time with re-input, because it usually ends up in chaos and wastes time with further corruption...

As an aside, I find OMR a very powerful tool which has helped me to capture over 60 scores (out of print and out of copyright).

Reply

• Oct 5, 2018 - 10:06

There's a Howto about fixing corruptions: https://musescore.org/en/node/54721
The easiest method probably is to select the corrupted measure(s) and swap voice 1 and 2 twice (and then remove the excess voice 2 rests). Works quite frequently. If not, there are more methods listed in that HowTo, including Ctrl+Del the measure (and sometimes the ovne bpre and after too), insert new measure(s) and reenter notes.

Reply

• Oct 5, 2018 - 14:53

I'm not sure what you mean about Measure Properties not showing the error. That dialog isn't there to show errors, it is there to show how many beats the measure should have. The error is if the actual number of beats in the measure does not match what is in Measure Properties. So in your score, Measure Properties is correctly showing you that bar 135 should have four beats. The corrupt is that it in fact has more than that, and that is why you get an error when opening the file. And the instructions described elsewhere in this thread are how you go about correcting such errors.

Reply

• Oct 5, 2018 - 19:34

Thank you DanielR and Jo-jo, for your comments and the referral to the "How To" on fixing corruption.
These are very useful in helping me understand how to detect and correct these errors.

Thanks also to Marc, for explaining better what Measure Properties does.
Up until Te Deum, I had been using SmartScore X2 Pro to convert printed scores to MuseScore files, for the benefit of my fellow community chorus singers. After a hundred or so scores, I've been able to narrow down the errors in recognition to those that Measure Properties detects as measures of incorrect length.

On Te Deum, I decided to try using PhotoScore because its interface seemed so much easier to manipulate.
It turns out that PhotoScore's error flagging process is different than SmartScore, and that measures PhotoScore showed as correctly timed, were in fact corrupt.

Thanks to the information provided above, I now know the proper way to deal with them.
My first step will be to use MuseScore's Corruption Details listing to return to PhotoScore and examine those measures, fix them there, and then re-export the XML file for MuseScore. This process will help train me to better spot PhotoScore recognition errors even when it has not flagged them as such. With such a process, I expect to eliminate the corruption of XML's as I have been accustomed to doing with SmartScore X2 Pro.

The initial error I saw when using MuseScore to open the XML from PhotoScore showed as a "fatal" error, yet I was able to "Ignore" it and open the file. Here's what I saw:

> File '/Users/.../Verdi - Te Deum from PhotoScore.xml' is not a valid MusicXML file
> Details: Fatal error: line 7022 column 16 Content of element accidental does not match its type definition: String content is not listed in the enumeration facet..

What is meant by "column" and "enumeration facets." These terms don't appear in the "How To Fix" page.
How might I find the source of this type of error so I can address it?

Reply

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