Condensed Scores, Thickened Lines and Part Extraction

• Sep 30, 2009 - 06:28

When I first started studying orchestration, I ran across a book on film scoring by Frank Skinner titled Underscore. As was common procedure, Frank scored films using three or four staffs, and handed the score over to his arranger, who did the grunt work of extracting the parts and filling in the missing bits and pieces. Instead of staffs being assigned by instrument, they were assigned by function. The breakdown was:

* Melody
* Countermelody
* Rhythmic support (chords)
* Bass

For example, when the melody was passed from the flute to the clarinet, the score simply indicated the change by writing clarinet in the staff.

This is no doubt obvious to all of you folk, but it was a complete revelation to me. Instead of being a confusing mess of dots, it was clear what each different instrument was doing. Since that point, I've been looking for music notation software that would support this sort of compositional approach.

Here's a portion of a condensed score from Henry Mancini's Sounds and Scores:

Here's a portion of the piano arrangement of the same song:

As is typical with much "pop" piano music arrangements, the piano right hand takes the melody, and harmonizes it with block chords. Having the software smart enough to do this seems to me to be a useful feature. So the "auto block voicing" would make sense, even without the first two features. (Band in a Box has such a feature).

So the set of features that I'm looking for are:

1. Dynamic reassignment of instruments to staffs (i.e.: flute+oboe, or oboe+flute/clarinet)
2. Dynamic reassignment of harmony voicings to a staff (i.e.: in thirds, octaves, close block)
2. Automatic extraction of parts;

It seems to me that having software that could support and automate the process of converting sketch into a full orchestration would be very useful to a subset of users. I don't recall seeing other music software that takes this approach.

Being able to reassign the instruments, and even the voicings on a sort of meta level would make orchestration much simpler than it currently is. For example, changing a line from 2 clarinets in octaves to 2 flutes in 6ths would be a matter of changing the instrumentation from:




and the harmony from:



6ths and 3rds

As a lazy orchestrator, the idea of having software do much of the grunt work appeals to me.

On the other hand, I'm not sure how good a fit this would be, since MuseScore is very much WYSIWYG, and this sort of approach is aimed more at composition than transcriptions.

Any thoughts?


Writing a short score (or condensed score, as you call it (in German and Swedish: Particell)) has been a common way of composing orchestral music since Schubert. The program Sibelius has a feature called Arrange that I think can do some of the tasks you are asking for. It would be interesting to read a report if you decide to dive into experimenting with that. Personally, however, I think for a notation program project, trying to perfect features like input is a lot more important.

In reply to by Magnus Johansson

I agree: Getting basic things like data input to work well is a much higher priority.

I've made various attempts to write something like this myself, but failed pretty badly.

Last year I ran across A Berklee Course on Arranging that someone had posted online, and decided to have a go at implementing it in Java. The program takes a subset of ABC notation, and outputs thickened lines, harmonic continuity, and so on. That encouraged me to start working on a UI for the program.

I had a look at canorus, but they don't appear to support chords yet. I looked at abc4j as a possible library for rendering, but there were a number of problems with rendering and - long story short - I was too stupid to figure out how to fix them. (There's been work done since then, so I should probably revisit it).

I started working on my own rendering library instead, but that's on hold after I ran across Noteflight. I realized I was putting all my focus onto rendering, when I should have been focusing on the user interaction and usability.

So now I'm again revisiting my assumptions. If I can find a program (like MuseScore) which can essentially do what my project's trying to do, there's no point in continuing to work on it. If there's nothing similar, I'll keep working on it, but I need to spend time designing the UI. But the focus will be a lot more specific - on quickly sketching a score exporting to MusicXML and MIDI, not as a general purpose notation program.

In reply to by [DELETED] 5

I'll have a look at it. I've got MusicXML export working more or less, but if there's already a tool that already does all the work, that's great!

I've got a bad habit of trying to re-invent the wheel. I learn a lot in the process, but I doubt the end product is better.

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