Something is wrong with lyric font kerning in 2.03

• Jun 29, 2016 - 02:55

Musescore 2.03 seems to add extra space with some letters in lyric text, that would make them difficult to read.

This is an example from 2.03


This is what it looks like in 2.02


Actually, the same characters look a little squashed in 2.02, just acceptably so. So I'm assuming you guys did a fix, but just bumped it a little too far and it can be easily fixed in next build.


In reply to by Shoichi

Exactly.... I looked at "Kerning — 1 step forward, 2 steps back"

Most decent fonts have kerning hints inside the font, maybe whoever is doing the kerning should use those hints. On professional fonts they actually work! I use Arial or Arial Unicode - deliberately for compatibility.

That is, they should use hinting on all operating systems! Mac used to be the king of this, before OS9.x, but kerning hints in TTF/OTF are much better than in the old days where if you wanted it to look right you had the option of PostScript or else PostScript.

In reply to by SeasonPsalt

As far as I know, kerning is handled by the Qt libraries we use. I don't have Arial on my Ubuntu system, but the kerning *does* basically work with other fonts I've tried. So perhaps the way Arial - a very old font - stores kerning info is not compatible with how Qt expects it - I know there are many different ways this information can be recorded in the file. If you have a font editor (eg, FontForge) you can try making copies of the font using different options for kerning to see if one works better for you. Or just switch to a more modern font.

In reply to by Marc Sabatella

I've got several font editors, professional ones, most of them older, but if the problem is with the font, why is the kerning acceptable in 2.02 and not 2.03?

I can try some other fonts, but Arial (though old) isn't static and has been tweaked for years. If I enter my own kerning info what might look great on my computer will look like junk on someone else's, even in a PDF if I embed it, and a printer stored font will override what I see on my screen.

What fonts does the new QT kern well? I know Geneva is probably history, and I've been out of the Apple world for too long.

Does QT prefer Postscript?

In reply to by SeasonPsalt

Different version of Qt, maybe they dropped support for the style of kern table used in Arial. Or maybe a bug developed. Anyhow I want suggesting you create your own kern info, just load the font then immediately save it but with different options for how the kern takes are stored (I seem to recall open type versus old style versus apple being possibilities).

The fonts we include with MuseScore - FreeSerif and FreeSans - should both kern well. I've never tried a postscript font with MuseScore.

In reply to by Marc Sabatella

FreeSans seems to work well with 2.03. I haven't checked to see what a generated PDF looks like on a computer without FreeSans if the font is not embedded. Does the website embed fonts when someone downloads "as PDF?" I'm guessing Emmentaler, etc is embedded.

Printer font overriding shouldn't be a problem with FreeSans, unless someone has their printer settings to force the printer to do that (usually done to make printing faster, and it's kind of old-school for general users).


In reply to by Jojo-Schmitz


Now that I'm looking at this critically, (after finding the "Common Fonts" ALL stink under MuseScore 2.03 with the new QT libraries,) FreeSans isn't that great, especially the italic if you alternate regular for odd and italic for even verse numbers. It really REALLY looks bad at the verse numbering, because of the italic angle of the glyphs, etc.

I tried a bunch of stuff, searching for "Readable Sans Serif."
I mean, a BUNCH of stuff. Hours worth. I already have gobs of professional fonts (mostly Adobe or Linotype, some from others) from various eras, but that would defeat the purpose of Freely available MuseScore to use these, so I didn't try any of them. I only tried freely available stuff.

The best looking Sans Serif I found freely available was Open Sans. I mean, it's so good you might consider replacing FreeSans with the regular (all 4 weights, Regular, Italic, Bold, Bold Italic) in future releases. It is as good as a professional font.

I'm not doing a "Let's have everybody on MuseScore tell their favorite font" deal with a gazillion people all telling you what you should have. This Font family is really good, and it's licensed under an Apache license. Of course I didn't read it thoroughly but It looks very similar to a creative commons license.

It can be also embedded into websites.

As far as Serifs go, I didn't look at any of them, unless you include Maestro Times, which is actually Roman. But users must have a MakeMusic product license, such as for Finale to use it. But it also looks great, except for their hideous hyphens.
But I'm sure MuseScore can't use it because of the license restrictions ;-)
I was actually tempted to alter the hyphens, rename the font, and use it for semi-personal use but thought better of it since the United States is a Litigocracy.
Those who think it's a Democracy or Oligarchy are misinformed.

Open Sans:


And this is what Maestro Times looks like. I LOVE it save for the hideous hyphens.


In reply to by SeasonPsalt

Maybe the height of the line MuseScore draws as a hyphen should be based on the x-height of the glyphs in a font, if the font was constructed correctly.

And thinking about it again, maybe new MuseScore revs should be compiled with the QT libraries that were used to compile 2.02 until their bug is fixed. People should be able to use standard fonts without kerning issues.

In reply to by SeasonPsalt

There has been quite a bit of discussion on how hyphens are or should be drawn. See for example #21310: Hyphen too high and #21239: Style of dash doesn't follow lyrics style. Part of the problem is that different fonts actually just sit differently relative to their x heights. But yes, we recognize there is room for improvement.

I don't know that we know for sure the issue is a bug with Qt - like I said, maybe they just dropped support for some obsolewte format. But certainly, each new versions of Qt introduces some bugs but fixes tons more. Going back to an older version means un-fixing all the bugs that the update fixed. Not worth it just to fix kerning on some particular proprietary font that we don't use by default and not all users will even have on their systems.

In reply to by Marc Sabatella

No, the kerning on the propitiatory font was fine!
I'm talking about kerning on STANDARD fonts that are STILL IN COMMON USE, even if the development world would love to call them obsolete, because they don't use or test on the platforms they are developing for (we'll say Windows, for example.)

For example, how can a developer of engraving software not have access to Arial?

In reply to by SeasonPsalt

As far as I know, Arial is a proprietary font - that's what I was referring to, anyhow. My point is, we can't unfix some unspecified number of bugs just because one one particular OS, one particular proprietary font doesn't kern well. The fact that the different OS's we support all provide different fonts and the libraries implement the font rendering primitives differently has been a sore point for years - scores created on one system will often look subtly different on other systems. This is why we moved to including text fonts with MuseScore and using them by default, and why we switched from Qt to FreeType for some (most? all?) of the font rendering we do (which is to say, I probably misspoke saying it was a change in Qt versions). It's more important to keep good and consistent rendering by default on all systenms than to allow this to break just to better support one particular proprietary font on one particular system.

In reply to by SeasonPsalt

Be sure to report this officially via the issue tracker, with sample score.

Be aware that by downgrading, you are trading this one one issue for litewrally *hundreds* of bugs that were fixed but you will now have to fight. Wouldn't it be better to simply use one of the fonts that kerns well rather than invite all those bugs back?

In reply to by Marc Sabatella

I'm not sure what I will be giving up with the downgrade, I've been pretty much happy with 2.02, my scores are pretty simple.

The only real annoying problem I have is verse numbering alignment if the first character in the verse is non-alphabetic, such as a quotation mark, similar to the issue reported here: which looks like you fixed.

But the quote issue didn't seem to be fixed in 2.03, I never really reported it.

You guys fixed the general verse number alignment issue by using control-space a long time ago, but it still doesn't like quotes as the first character in a verse.


In reply to by SeasonPsalt

Here is a list of some of the hundreds of bugs fixed between 2.0.2 and 2.0.3:…. Quite a few very serious bugs fixed, including some that could crash the program, leave your score corrupt, etc. Plus some very useful new features you'd be passing up on.

Feel free to formally report the issue you are seieng with quotes. I can tell you that depending on exactly what the context is, sometimes you might want a quote ignored, other times not, so it's pretty much impossible always do exactly what the user might want by default. Different users might want different things at different tiems for different reasons. So there is no getting around the occasional need for manual adjustments.

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