Fonts for music typesetting

venticaratteruzzi's picture

Any suggestion for music fonts from good foundries?

Andreas Stötzner's picture


Could you be more precise on the task in question?

Frode Bo Helland's picture

I’ve heard about music type from the metal days but nothing (of high typographic quality) recent. I’ve been dreaming about picking up the tradition for a while, but there’s always another more important bezier to draw. Do you have a budget? Interested in collaborating on something?

Andreas Stötzner's picture

The problem is not drawing the Glyphs. The problem is the encoding. There are at least three (!) completely different ways of encoding musical notation out there.

Have your pick :-)

Grzegorz Rolek's picture

Except for an excellent and well researched Caeciliae or Gregoria, both for Gregorian chant, there’s nothing decent really. P22 got once a Music Pro package with modern notation, but I see it as an experiment for nothing more than rapid typesetting of a simplified double staff notation only.

I have serious plans for making some real revivals of mensural music types like those ascribed to Jacomo Ungaro, Petrucci’s type provider. I’m also thinking to make Caeciliae, as it’s open source, to be Unicode compliant and its input more suited for professional typesetting environment, not just word processor. Anyway, it’s good to hear someone's picking those issues up.

Birdseeding's picture

Most music notation software/scorewriters (the big ones being Finale and Sibelius) have their own proprietary built-in fonts. That's also the software most music typographers ("engravers" is the official but technically-incorrect term for some bizarre reason) use, and I know at least Finale is able to import custom fonts.

The fonts included in my copy of Finale are pretty-awful-to-deeply-abysmal, and better ones would definitely be appreciated. I do think separate versions would have to be made for the two big software players as they probably use different glyph sets with any luck.

Andreas Stötzner's picture

…revivals of mensural music types like those ascribed to Jacomo Ungaro, Petrucci’s type provider

A good idea. I happen to have done it already, halfway through. The question is, is there a market for such fonts? Is there a market for re-typed Petrucci editions? And again: there is no encoding yet for mensural notation characters.

The fonts included in my copy of Finale are pretty-awful-to-deeply-abysmal, and better ones would definitely be appreciated.

It seems to be similar with Sibelius. I know musicians from the Leipzig Gewandhaus Orchestra who are becoming sick while playing year by year from badly “engraved” music. In the opera it is dark and they just don’t see what they are supposed to perform. They have now decided to start making their own editions because the major music providers obviously ignore the problem.

hrant's picture

It's strange that it's not in Unicode.


Birdseeding's picture

It's strange that it's not in Unicode.

Anyone have a good answer to this? Even if some symbols are sort of unsuitable to encode as straightforward glyphs under the current system (because they need to seamlessly and without distortion stretch to any length), a lot of them are very clearly modular or standalone and all of them are typographic and/or chirographic in nature. Also, the usefulness of having many of them exist in running text is demonstrated by the "Bach font" above.

Even the most complex of the fonts included with the music typesetting software has no more than maybe 200 glyphs. And a lot of them are awesome typographic-origin symbols that I hope type designers would enjoy drawing. ;)

Grzegorz Rolek's picture

Is there a market for re-typed Petrucci editions?

It yet has to be created. Modern transcriptions lose information or don't exist at all. More and more performers or scholars would like to or thus are forced to read the source notation, but contemporary editions of original notation are a true rarity. Facsimiles, in turn, in today's digital diversity are becoming frustrating to use. Not to mention there's no way to work with such documents textually. Besides, there's clearly an emerging market for encoded modern notation; the early one, I think, will follow.

…there is no encoding yet for mensural notation characters.

But there is, right in Unicode. Mensural notes, rests, mensuration signs, croix. For other characters like different clefs, repeats, flat, signum congruentiae or corona there're code points of their modern counterparts, at least as I understand it, just like a uncial script shares its code points with a humanist. I think there's only custos which is missing, but it had no real textual function, so I can imagine Perry Roland, who proposed the encoding, decided just to omit it entirely (unfortunately, in my view).

hrant's picture

> decided just to omit it entirely (unfortunately, in my view).

Note that submitting a proposal for an addition
to Unicode isn't too daunting. I'd go for it.


Andreas Stötzner's picture

It's strange that it's not in Unicode.

No it is not strange. Unicode is not the Holy Bible. It’s the result of a process, conducted by humans, influenced by interests.
Musical notation is a very, *very* complex and disparate subject (in technical terms).
It is perhaps even more complicated than Mathematic formula composing or Egyptian Hieroglyphs. To encode that to full extend, this requires considerable effort and research. Who is supposed to fund that?

Yes there are a couple of musical ‘symbols’ (stupid term) in the UCS. One can use them for putting tiny pieces of notation into running text, that has its uses, surely. But you can not set actual music scores with it. If you look at Petrucci’s lettercase: there you have the *character set* needed for that – and that is not encoded.

It is similar with Byzantine musical notation, neumes, square notation, mensural notation and modern notation: a few basic glyphs are encoded, but that is not sufficient for setting actual music scores.

Now the interesting point with the lead-type based mensural notation (as introduced by Petrucci) is: it is technically the only *discrete* system of music notation. Because all other habits of doing it where either handwriting or engraving-like, hence not a system of limited discrete parts as in the lettercase. So, for that reason, Petrucci ought to be the actual first candidate of music becoming encoded at all. Because this system is more related to our system of text processing than other kinds of music.

Grzegorz Rolek's picture

Hrant, how many years it took capital eszett, with support of the community and later the DIN itself, to be included in Unicode? Four? (Hat tip to Andreas, by the way.) But even that is not the real issue. Problem with any changes to such things as music notation characters in Unicode is that there hasn't been yet any real, large-scale applications of it (no I'm aware of) and it's hard to tell what really works and what doesn’t.

hrant's picture

> Who is supposed to fund that?

The guys who funded Klingon? :-/
I'm kidding of course, but I'm not kidding when
it comes to remaining surprised that something
as central to human society as music hasn't been
properly encoded yet. Unicode has been around
for over 20 years, and music has been around for
thousands of years; people spend a ton of money
making, nurturing and listening to music. Collecting
money to see this realized should be a piece of cake.
There must be something else going wrong. Might it
be that musicians are bickering so much over how
to do it that nothing is getting done? Just a hunch.


Andreas Stötzner's picture

Hrant, in principle I agree with you.

You’re talking about *what*, that is idealism.

In the real world, things happen according to the *who*, that is realism.

There is no ‘High Council of Human Script and Notation’ which would wisely conduct things which are reasonable. Unicode is funded by MS, Google, Apple and all the other big players who have their interests. Petrucci did not belong to it so far. Nor did Beethoven, I’m afraid.

Look at another example: the big players force sets of “symbols/emoji/pictographs” into the standard. Which is actually a worthy task, but: HOW they do it. (Can’t tell you how the recent picto additions would have looked like if not Ireland and Germany would have been so lunatic to fight against it like mad idiots. Smashing “Windings” into the standard because it’s declared being “popular”, that is the ratio behind those endeavours. Little thought, little reflection, little intelligence; no science, no principles, no scruples. Nothing.)

Despite the issues mentioned I maintain that Unicode is a wonderful thing.

I just point at this general problem briefly, to prevent you from assuming that Unicode would be build on grounds of any reasonable blueprint. It is business interests and bloody pragmatism, not more, not less. If you can deal with that, you may succeed. But be prepared for the worst.

But back to music. I’m interested in that because I’m working on a long-term editorial project in that field. During the last 14 months I have been typesetting a volume about the musical notation (Europe) of the 17th and 18th century. Of course I use the relevant established codepoints for eminent musical characters which occur in the text. But, the whole set I actually need for composing that treatise is four times as large as covered by te UCS. So it goes the usual way: private encoding of special characters. It is a hack, and everyone else will remain hacking for himself as long as there is no substantial research and development initiative for music encoding, possibly focussed on feeding the ISO/UC encoding procedure. Taken the particular complexity and diversity of the matter into account, I don’t see any possible stakeholder to embark on this so far. – Maybe this is a general issue: the week position of the humanities before the computer industry. We will not solve this problem tonight.

Just proposing single additional characters may help in some respect, but it will not solve the overall conundrum.

What are YOU listening to??
Me: Crystal Radio Beautiful Music. “All Day, all night, all nice” :-) very cheerful.

hrant's picture

> It is business interests and bloody pragmatism, not more, not less.

Sadly, I believe you.

> What are YOU listening to??

As much as possible, Goa - which I consider
to be the Classical of the 21st century.


Grzegorz Rolek's picture

Unicode’s musical symbols were indeed intended for running text or to be used with a markup for layout, so technically it's not suitable for music scores, but Petrucci’s settings were no scores either. The very nature of Petrucci setting his editions was just a set of isolated lines of music without any special spacing or lyrics underneath. Invention of white mensural notation allowed that and that’s what Petrucci took advantage of, hence similarities with a running text.

This doesn't mean earlier or later music notation wasn't discrete. In fact, early music incunabula were mostly a rather successful plainchant settings of discrete metal types. It just has had to be spaced along underlying text, the very basis of plainchant. (In fact, given the inseparability of text and neumes in plainchant, the latter could be also treated as some form of combining marks.)

Now, a modern score is just a few more such parallel runs of music lines. Proper spacing would still be needed for horizontal synchronization, but it could be automatically derived from the text itself. What really is missing in the Unicode for music to be encoded fully, then, are control code points for vertical position of characters within any such line, i.e., within a staff. That’s it.

Yes, there are exceptions like with the fermata or dynamics, which in such case should be encoded also as non-spacing combining marks. But this, for now, could be bypassed by treating it as another such parallel line of text, just as LilyPond software does it conceptually. And sure there will always be some peculiarities of notation in any given era or site not yet encoded or encoded badly, but that's the case with most non-mainstream scripts.

So, Unicode's not perfect, no doubt for that, but the beast called musical notation isn't so complex as it seems, and things with the encoding don't look as bad as that.

Grzegorz Rolek's picture

I should probably add that the flexible spacing in modern scores, which maybe gives it this overall complex appearance, is just a visual aid and is by no means subject to encoding. Scores can be easily disassembled into isolated lines, all the spacing taken off, and there would be no loss of information. It still would be read as properly synchronized, running texts in parallel. In fact, this is exactly what is being done with a full score while creating sheets of individual parts for individual performers to read.

Jens Kutilek's picture

Incidentally I found an illustration from the old days of hand setting:

The bottom part is a synopsis of the »around 360 type elements necessary for music typesetting«. The upper right hand part I find particularly interesting, it shows the separate type elements that combined produce the image below it.

The accompanying text notes that this technique was only be feasible for short passages of music that had to be printed together with running text in a book. For settings which consist mainly of music notation, the text explains that the music would usually be engraved, printed and then an etching would be made of it.

The book is »Buchgewerbliches Hilfsbuch« by Otto Säuberlich, Verlag Oscar Brandstetter, Leipzig 1938.

eliason's picture

The upper right hand part I find particularly interesting, it shows the separate type elements that combined produce the image below it.

There's the recipe for an OpenType solution if anyone is cooking!

hrant's picture

Awesome find Jens.


JamesM's picture

The comment about hand setting reminded me that Letraset used to sell music symbols as dry transfer letters (on a clear plastic carrier sheet and you'd transfer them to paper by rubbing with a stylus).

Not practical for large projects, but I remember using it years ago for some small music projects.

hrant's picture

Seeing the elements they used would be useful.
Any chance of digging anything up?


JamesM's picture

Sorry, I tossed away my Letraset sheets many years ago.

de l'education's picture

some small example of dry transfer here
in the same site, some info about old ways to print music notation (the site is not complete yet).
I even heard about music linotypes (i really would like to see one of them!!)
anyway, the best resource about all the history of music printing is this book by Kummel & Sadie
music printing and publishing

Jens, I can't see your image... are you talking about mosaic type? P.S. Fournier wrote a lot about it in his manuel typographyque.

about modern music fonts, the market is linked to specific softwares (i know about rare music foundries for music fonts only, they use to set the fonts for softwares like Finale or Sibelius).

One of the most interesting works of type-design I know was made for the Lilypond font, you can read an essay here

Andreas Stötzner's picture

The book I mentioned earlier in this conversation has been released these days. For those who are interested:
here are some views and further links.

quadibloc's picture

The problem is not putting the symbols that occur within a musical score in Unicode. That, as has been noted, has been done, to some extent.

The problem is that a musical score is more than a string of characters. If one thinks in terms of a left-to-right writing direction, then how is one to indicate which notes are to be in the same vertical column as which other notes, indicating that they are both sounded at the same moment in time?

Essentially, music scores are two-dimensional. Each note belongs to a certain part, and begins at a certain moment in time. ASCII and other character codings are inherently one-dimensional. The problems this causes for mathematical notation are basically solved by treating the notation as a series of nested brackets for constructs such as "superscript/subscript pair". Musical scores need a completely different approach.

Which is not to say that there may not already be an elegant solution involving TEX, but sadly, despite its power and elegance, that system is mainly relegated to the sidelines for a number of reasons.

There are open-source programs out there which generate musical scores, I believe - I know there are some that are at least freeware. Thus, there should eventually be an accessible encoding that handles the features required for specifying a musical score of reasonable appearance.

Té Rowan's picture

As far as I can recall, the note editor Lilypond is open source. I think it even has some reasonable fonts and faces.

HVB's picture

Té: Lilypond comes with 9 overlapping otf font files named emmentaler-11, -13, etc. They have extensive music notation symbols in the private use area.
Also included are four URW++ Century Schoolbook fonts with added Cyrillic - unusual for a GNU product, to say the least!

Té Rowan's picture

Oi dunno... As far as I know, the GPLed urw++ fonts have all been Cyrillised by one Valek Filippov.

The main driving force behind many open-source projects is 'scratching my itch', and if it scratches someone else's itch, too... why, dat's a bunce.

John Hudson's picture

Unicode is unlikely to encode characters for music notation until something like Music XML becomes an official standard. A parallel is the development of MathML and subsequent encoding of math alphanumeric characters. Encoding specialised characters without first establishing and understanding the framework in which they will be used is not a good idea.

Theunis de Jong's picture

That comparison with MathML is excellent. You cannot create "a math font" for pretty much the same reasons. An equation may have superscripts, and in those there may be subscripts again. And they, in turn "have smaller still to bite'em". What would a font for that look like -- every character in every possible smaller size? The same goes, of course, for horizontal and vertical extensions of braces and division bars.

Just as a "regular" OpenType library such as UniScribe handles the logic between font and output, this needs a combination of font plus a layer of software to handle it.

thetophus's picture


As a musician I'm interested in anything that can help me quickly notate music. I'm not smart enough to write scores (I play rock and roll) but I'll take a robust notation system.

jcrippen's picture

Traditionally most music was directly engraved on metal plates. This was true through essentially all of the 20th century despite all the technological improvements in other forms of paper publishing. It’s only recently that computer-engraved music has started to catch on, and even then a lot of professional orchestral musicians dislike it (with good reason). YouTube has a video or two from G. Henle demonstrating the process.

The attempts at typesetting music were never very popular and never produced anything more than tolerable results. Music is much more tricky than text. It’s at least equal to the complexity of typesetting mathematics, if not more difficult. For starters, music is inherently two dimensional in comparison to the linearity of text. Whitespace is also radically different, and fairly minor changes in spacing can have profound effects on the way we read music. Music also often has to fit a grid, at the very least in bar structure, whereas text ideally should not be vertically aligned (otherwise we see rivers etc.).

Lilypond is probably the best computer engraving software that is publicly available. (There are probably in-house systems in use by music publishers which aren’t available to the public.) The default text produced by Lilypond leaves something to be desired. But the music it produces is quite good and often approaches hand-engraved quality. Lilypond is also constantly improving and the developers benchmark their work against hand-engraved examples. It’s really a major triumph for free software, since Lilypond has essentially surpassed all the commercial competition, at least what’s available outside the walls of music publishing companies.

BTW, the term ‘score’ for printed music comes from the scoring of the five lines of the staff into the metal. That’s the very first thing in physical engraving of music.

Grzegorz Rolek's picture

…a lot of them [musical symbols] are awesome typographic-origin symbols that I hope type designers would enjoy drawing.

Emilio just picked this up: music notation in style of Didot, Garamond, and more as a thesis of Giovanni Murolo, Nuovi metodi di rappresentazione nella notation musicale. See the pics in the gallery or at Flickr. The process of adaptation he has used seems rather naïve, but these shapes look really nice and certainly were a joy to draw.

ahyangyi's picture

@Grzegorz Rolek,

That's amazing and inspiring!

Liuscorne's picture

This might be of interest. It's part of the development of a new music notation program at Steinberg.

Grzegorz Rolek's picture

Here’s another nice take on music typesetting: Notationstypografie, a thesis by Stephanie Türck. These are six different works from different contemporary composers — all set with different, custom type. For each work there’s a dedicated page with detailed photos of the score in print. They look great.

charles ellertson's picture

I kinda skimmed this thread, but there are musical symbols in plane 1 Unicode. Didn't see any mention of this, or perhaps I missed it. I do remember mapping the Adobe Postscript musical symbols font (Sonata) there when things went from *names* to Unicode...

In any case, as John said, the problem isn't just with the symbols per se, but as with mathematical symbols, the necessary software to set music.

Syndicate content Syndicate content