If you read sheet music, you may have wondered how it's
produced. Are there armies of people frantically pointing-and-clicking
with Windows score-editing packages? Do they rush out and buy the
latest version of Finale whenever an upgrade comes out? You can imagine
how to print hundreds of thousands of copies of the latest Harry
Potter: it's all plain text, and it's easy to see that you could use
some standard software to typeset it. But for a complex score, with
less money behind it, in a smaller print run?
The answer, surprisingly, may involve Linux. Here
Linux Musician presents an interview with Mike Mack Smith,
co-founder of Barnes Music Engraving, one of the UK's most
well-established and respected computer-based music typesetting
companies. Interview by Chris Cannam.
Chris: So first, why "music engraving"?
Mike: It's an historical thing, dating from the time that music was
engraved into pewter plates, in mirror image. We use the term
We justify it by the fact that the image is "engraved" onto
photographic paper by a laser, but music typesetting might be a better
What form do you receive the music in?
Usually we receive a manuscript from the publisher, maybe hand
written, or a previously set work with alterations, or set by another
computer based system. Or for some publishers we also act as
arrangers: they supply us with a CD and ask us to make an arrangement,
say, for piano and vocal with guitar chord boxes.
And what form do you turn it into?
Until about four years ago, we used to supply the final image from
a high resolution imagesetter, usually around 3,000 dpi. Now, all
finals are supplied as PDF files. PDF seems to be a pretty good
standard. We used to have awful problems when we supplied stuff in EPS
form. There were often many problems sending files to printers,
especially if they were Mac based -- usually to do with embedded
fonts. So far there have been no problems with PDF, and they are so
compact that delivery via e-mail works very well.
Can you give a quick description of the software and
hardware setup you use?
A basic Linux installation. We've used SuSE for the last couple of
years. The music typesetting system is called Amadeus, developed by
Wolfgang Hamann and Kurt Maas and first released about 17 years ago,
running on PDP11/73 with the Idris operating system, which was a Unix
lookalike. Input was on terminal monitors. You didn't see the music
until it came out of the imagesetter. Strangely, this was very
good training for everyone!
Around 1990, we started using Atari STs, and we could actually see
the music on screen. And in about 1994 started using Linux. Wow!
The software is really a music formatting system. You give the
basic music information for each stave, with no concept of
layout. Then when you have all the parts, you give some dimensions and
... you have proper music! It's very un-point and click. We
can now enter whole lines of music without looking at the screen once,
just at the manuscript -- although in practice we do check what we are
typing. It's basically copy typing for music.
It combines well with text formatting -- a version of troff -- and
with programs for incorporating TIFF images etc, and of course it also
generates midi files. Wolfgang Hamann has had to do some re-writes on
software, and we still have perennial problems that the software still
uses some very old libraries which SuSE stopped shipping after
The reason we stick with it, rather than something like Finale or
Sibelius, is that we can customize it to do whatever we want, not what
a programmer has determined in advance that we want!
It sounds a bit like Lilypond. Have you ever
considered trying to switch to a free-software alternative like that?
Any idea what the pros and cons might be?
Well, we believe that Amadeus produces the best quality in the most
efficient way. Music publishers can be very picky about the smallest
aspect of music setting... and so are we. Certain music
setting packages may be great as composing and arranging tools, but do
have deficiencies when it comes to the presentation of the music for
I haven't had a long look at Lilypond, but my guess is that Amadeus
is more advanced. It can, pretty much, deal with anything, so long as
it's conventional notation (and some deviations).
The cost of any software or equipment doesn't figure highly: we'll
buy anything that enables us to produce our product more
efficiently. When you're doing it eight hours a day, any improvement
in efficiency will soon pay off.
And the rest of your setup?
System wise, on site we have a server and backup server, about five
Linux workstations and three Windows workstations, and printers for
supplying proofs and check proofs. We need the Windows stations for
company financial software. We can't find appropriate stuff in Linux
-- Sage e-mailed that they had no plans for a Linux port. We also tend
to use Photoshop, although I'm trying to force myself to use Gimp. The
most important use for Windows is to run Acrobat Distiller, which
seems to always produce a bomb proof result. We have had problems
with the Postscript to PDF converters within Linux.
Sometimes we also need to use page makeup programs such as Quark or
Adobe InDesign to combine graphics, music and text. Sometimes it's so
complex you need to see on screen what's happening, and be able to
Do you have a typical sort of client with a typical
style of music?
We'll work for anyone! ... Actually we won't. We tend to work for
established publishers. Working for private individuals never seems to
work well. We work for publishers such as Oxford University Press,
ABRSM Publishing -- the people that do all those exams --, Novello,
Boosey & Hawkes.
Types of work are hymn books, full orchestral scores, choral leaflets
and anthologies, educational books with text and music, exam papers,
piano, vocal and guitar box arrangements of pop songs, easy keyboard
And are any of these particularly difficult to deal
Each of them provide their own particular problems. Full scores are
probably the trickiest, and least profitable. Hymn books are usually
a case of organization: there can be over eight hundred items, which
have to be set in a coherent and consistent style. Educational books
can be a bit of a pain as we have to use Windows for page makeup
programs, Adobe InDesign or Quark Xpress.
Composers sometimes start inventing their own non-standard system
of notation. This is not good news for anyone! We will usually drop
these projects as soon as possible!
So how much involvement does the client get in deciding
how something will look, and how much control do you get?
It very much depends on the customer. In some cases the customer
leaves everything up to us; this is sometimes a good thing and
sometimes a bad thing! When it's done by the customer knowing that we
are able to take what they have given us and create a good product,
it's good. When it's done out of ignorance by the customer, it can be
We will usually send samples back to the client before we set the
whole job, for them to okay, and any alterations to style from
that point would be chargeable.
How much work goes into typesetting a single piece?
Pages can take anything from ten minutes to half a day to set. It
depends on the complexity of the music, but also of the music
typesetting software's ability to deal with it. We usually spend quite
a lot of time setting up a job -- centralizing data as much as
possible, making shortcuts, etc. A good setup greatly reduces the
amount of time an individual page takes to set. For a hymn book, we
may take a week or two just doing the setup. This always pays off.
Don't you get sick of seeing and hearing the same thing
over and over again?
Surprisingly, all pieces are different, providing their own
problems. Sometimes it does get a bit like a production line, but
then we just try to get it done as efficiently as possible -- "what
the heck, where's the cheque!". This tends to be quite an incentive!
Listening is only a small part of what we do. After the piece has
been set, we will give it a quick MIDI proof, just to check that there
aren't any major blunders like wrong key signatures. The most important
proof is the side-by-side visual proof, comparing the proof against
the original manuscript.
Whilst we're setting the music it is very much a visual thing: we
don't consider how it sounds, only how we can lay it out to help the
performing musician as much as possible. After having worked on a
piece of music for a few hours, it's usually quite interesting to hear
what it sounds like -- and if it was worth it!
Is this the sort of business you can imagine readers of
this article being able to get into, if they wanted to? How would
someone interested in this set about finding work doing this sort of
It's an overcrowded pool already.
Ten years ago, life was much easier. Then along came these packages
that you could buy for five hundred quid, as opposed to ten grand for
Amadeus, and everyone who couldn't get a job after leaving music
college went into competition with us. We did warn publishers that
there was more to it than just having a bit of kit -- but back then,
music publishers were run by accountants, and it took a few years for
the penny to drop. After getting stung quite a few times, publishers
have learnt: single operators are used, but only when they have a
proven track record and have a body of work to show.
Prices have tended to remain static for the last ten years. The
only way we can survive is to constantly get more efficient.
Finally: the name of the company is Barnes Music
Engraving Ltd, but who or what is or was Barnes?
When setting up the company sixteen years ago we needed a name. We
were trying to think of something that gave the impression of
stability, avoiding the fly-by-night impression of "Lasermusic", or
"Cadmus" or "Jet Set Music". After going through various local Sussex
connections such as "Dowlands" or "Woolfs" we ended up going through a
list of people we were at school with. After discounting "Nibble" and
"Oraface", my collegue suggested "Barnes", which I thought had the
right tone. We even invented a founder of our company called "Malcolm
Earnest Barnes", who still has an account on the Linux system!
Oh well, it seemed funny at the time!