.musicxml vs .mxl

[motet]

Is there any reason to use uncompressed vs compressed MusicXML? Can Dorico and Sibelius read either?

[michelp]

Michael Good : "I'd recommend saving compressed MusicXML files (.mxl) and making sure that "Include linked parts in compressed files" is checked if you use linked parts. That way you save both images and linked parts which go missing in uncompressed MusicXML files. I'm not sure if Dorico imports those yet, but you'll be ready for future versions that might."

Daniel Speadbury (Dorico) : "When exporting from Finale, it’s a good idea to use the .mxl compressed MusicXML format... because it will (in Finale 27) include additional information about instrumental parts as well as the full score.
Dorico doesn’t currently import the formatting information from parts, but it’s something we may well support in future, so it’s a good idea to export the most complete MusicXML files you can.
Finale 27 will stay at version 4.0. But version 4.0 is a mature and complete format, especially for the purposes of moving between programs like Finale, Dorico, and Sibelius."

[motet]

Thanks.

I noticed a couple of things MusicXML doesn't do right off the bat: resized notes in cues, and the size of bounding rectangles around rehearsal marks.

I have a whole bunch of pre-Finale27 files to convert, but want to use Finale 27 since it has the latest MusicXML. Sometimes when you open a pre-27 file in 27 it asks, "Do you want Finale to mess with your articulations and make a mess of them in order to take advantage of the not-very-useful new stacked articulations feature?" (Not those exact words). Does anyone know how the "batch convert a folder to XML" utility answers that question?

[mmike]

Does anyone know how the "batch convert a folder to XML" utility answers that question?

Exporting a whole folder (including subdirectories) with the batch xml (compressed) command, no questions were asked, all were exported (inlcuding parts) and could be imported (opening as "untitled)

[motet]

I'm batch-converting a folder of Finale 2014.5 files with Finale 27. When you open an older file with Finale 27, it sometimes asks a question about conversion. Since the batch process doesn't ask any questions, I'm wondering what it assumes as the answer to that question. Looks like some experimentation is in order.

[mmike]

I vaguely remember when installing the latest Finale version that there was a choice offered right away whether to want Finale to mess with your articulations, etc., which I did NOT ever want to happen and therefore I declined that option. And that's probaly (I think) I never see that message again when opening any old files. So I assume that the batch process takes that into consideration. But you might of course want to test this with a few old files.

[motet]

I will do that and report.

I vaguely remembered a discussion about the articulations that involved going through them all and unchecking the "stack" flag, and something to the effect that the MakeMusic crew had been in a quandary about doing something backwardly-compatible but had decided to go ahead anyway. But I never really understood it, and that was the point where I decided to stick with 2014.5 for the duration. This XML excursion is mainly to have a lifeboat. I suspect using it to switch to Dorico or whatever would be a major project anyway, and that the articulations would be the least of my worries.

[mmike]

I suspect using it to switch to Dorico or whatever would be a major project anyway, and that the articulations would be the least of my worries.

Exactly ...

[motet]

I have a lot of pieces made with Finale 2005b and Finale 2011 and I've kept those versions installed all these years to occasionally fix little problems. As part of this MusicXML process I'm converting those old pieces to Finale 2014.5 first and comparing results page by page with the old one (I interleave the old and new PDFs and then go through and flip back and forth between pages, which makes it fairly easy). I was pleased to discover that, so far at least, the only problems I've found is that expressions which have been manually adjusted are occasionally not right. Slurs are often slightly different, but almost always for the better. I think I still need to do this page-by-page comparison to be safe, which will be time-consuming for 20-odd operas, but am relieved that there's not much to fix. So my goal is to have everything in Finale 2014.5 format and to use Finale 27 to generate MusicXML from the 2014.5 versions. Then I only need to keep Finale 2014.5 running. We'll see who dies first, Finale 2014.5 or me.

[mmike]

Ae you sure you need that extra step (old file > 2014.5 > 27)? I looked at an old file in 27.4 (created in 2000.5, modified in 2009.2.3), and it looks perfect. I then created an xml export and imported it again, and it looks basically the same with the exception of some non-standard graphical elements. Just a thought.

(P.S.: I'd vote for 2014.5 dying first ...)

[motet]

Finale 2014.5 is my working version, so I'm converting everything to that. I've found occasional differences between 2005 and 2014.5, and between 2011 and 2014.5, that I've had to correct, and I need to carefully check. Finale 27 doesn't work well on my machine (stuck keys), so I'm only using it for creating MusicXML since it has the latest version.

Regarding the articulation-conversion question, there's a "do this always" box ("No," in my case), which I've checked now.

I've also decided to save compressed MusicXML only, since it stores linked parts. In reality, it's a .zip file and you can get at the uncompressed score and parts if you need to.

I hope you're wrong about 2014.5 dying first, but if it does, then Finale 27 will open those files.

Has anyone tried importing a .mxl file with linked parts? Are they still linked?

[motet]

In a few cases, slurs have changed for the worse from 2005:

Finale 2005:

Finale 27:

(If I re-enter the slurs, same result; they must be hand-corrected).

[John Ruggero]

Strange. As I write the response, the two examples in your post shown below the "post a reply" box switch places and the Finale 2005 becomes the Finale 27 one. In any case, I think that the one that clears the augmentation dot by a mile is worse than the other in shape. The one that crowds the dot is a better starting point for some hand adjustment. But maybe the settings are different in the two versions and adjusting those would take care of it?

[motet]

I see what you mean about the images switching! Gotta be an error with the forums software.

At this point all I can spare the time for are quick adjustments. Here's what I did:

[motet]

Here are the slur options between the two versions. They seem about the same 2005b is the one that lets you specify angles to the millionth of a degree.

[John Ruggero]

Here I my settings, which have some differences:

Despite the differences, I get the same result as you with Finale 25.5 and seen at A.

Steep slurs.png (5.75 KiB) Viewed 1759 times

B is the hand-adjusted version.

The problem is mostly the large interval and resulting steepness of the two-note slur. No software that I am aware of handles this situation well, because the tips of the slur really should be shifted to aim at the sides of the note heads rather than the center. A couple of mouse clicks to the right generally does the trick. (Edit I should have said a couple of right arrow key clicks.)

But then there is augmentation dot which makes the situation even worse...

[motet]

Thanks. I ended up doing this:

[John Ruggero]

You are very welcome, motet. That looks great!

[motet]

The differences I'm finding in converting old files are mainly slurs--changed shapes, but also sometimes they're missing, or inexplicably flipped the other way--but also an occasional missing or extra cautionary accidental, and even a couple of changed notes. Imagine if you upgraded your word processor program and it changed some of the letters and punctuation!

[BuonTempi]

MusicXML was always billed as a 'lossless' format, in which everything -- content, layout, style -- was retained, particularly during the phase when MM responded to any problem in a Finale document with "export it as XML and import it back". But it never completely lived up to this, and the format is sufficiently expressive that different apps can interpret it in different ways.

There are other XML or text-based music description languages, such as MEI (which preceded MusicXML); and MNX, which explicitly claims to "avoid the design choices of MusicXML". Members of the MNX W3C committee include (...check notes...) Daniel Spreadbury, Michael Good, and Robert Patterson!

[motet]

The differences I describe have nothing to do with MusicXML. I'm converting old Finale 2005b and Finale 2011 files to Finale 2014.5 saving MusicXML. I'm comparing PDFs generated by Finale 2011 with those generated by 2014.5, for example.

I don't expect MusicXML to be lossless and it's definitely not. It's just a lifeboat for when Finale stops working.

[motet]

I noticed when converting some files made with Finale 2011 to Finale 2014.5 that 2011 did not kern text! As you can see in the picture below where I've drawn a red line, the "o" is not tucked under the "V" in the 2011 version. This is true both in the screen display and in the generated PDF files.

Could this be the font annotation business, which I've never investigated? I've never touched the font annotation, but I compared the Times New Roman V's setting between the two versions and they appear to be the same. I always assumed Windows was supplying the character spacing anyway, and indeed, MS Word and my paint program both match 2014.5's kerned version of that word "Voice" exactly.

I also have Finale 2005b, which kerns properly. So either it's font annotation/some other setting, or else 2011 (or some other version between 2005 and 2011) broke the kerning, which they later fixed.

Edit: compared the Times New Roman font annotation files for 2005b, 2011, and 2014.5, and they're all identical.

[motet]

And the non-kerning occurs regardless of fonts.