This repository contains digital music scores in the Humdrum data format from the Josquin Research Project. The primary web interface for these scores is http://josquin.stanford.edu which allows online searching and browsing, conversions into other data formats, such as MIDI and graphical notation, as well as interfaces to some online analysis tools. These scores encompass music of the early Renaissance (ca. 1420 to ca. 1520), mostly representing the Franco-Flemish School and are mainly intended for computational analysis. As a result only a few works include text/lyrics suitable for vocal performance.
A peculiar aspect in music of the sixteenth century (from a modern viewpoint) is an incomplete notation of accidentals needed for performance. Editorial accidentals have been added to these scores to realize the performance practice of the time. An "i" character immediately following an accidental indicates that it was not written in the original notation. When written without an "i" after it, the accidental is found in the original notation, either from the key signature, or an explicit accidental written immediately before the note.
Another conceptual difference is the use of mensuration signs. These evolved (decayed) into modern time signatures. The most common mensuration in the music is Cut-C, from which modern cut-time originates. In this mensuration the beat is at the whole-note rhythmic level, and the metric cycle is two beats (a breve). Circle mensuration is the second most common mensuration in the music. This represents three whole-note beats per rhythmic cycle (dotted/perfect breve). Mensuration signs also convey tempo, so MIDI files on the JRP website made from these scores typically interpret tempo from the mensuration signs in the data (with the tempo of the whole notes slowing down over the generations due to smaller rhythmic values becoming more common). Modern time signatures are present in the data as a result of using music notation software for data entry and are of secondary concern (notice that they are not displayed in the scores on the JRP website, but use rather for metric analyses.
Barlines had not yet been invented in the fifteenth century, so all barlines in the data are interpreted; however, these are less open to interpretation compared to editorial accidentals. Typically barlines represent breves (double whole notes) durations, although they sometimes represent longs (quadruple whole notes), and may include patterns of dashed and solid lines to indicate the implied long (solid lines) and breve (dash line) metric grid of the music. In Circle mensuration the mensural breve is a dotted double whole note in modern notation.
All parts are encoded in modern notation, but sensitive to the
original mensural notation intent of the fifteenth century. The
main difference is that all rhythmic durations and augmentation
dots are made explicit rather than dependent on a note's musical
context. Since the original parts were not written in modern score
layout during the fifteenth century (see an example of a
partbook),
mensuration changes sometimes do not occur at the same point in
each part. For example, one part may have a whole note in Cut-C
mensuration while another part has a dotted whole note in 3
mensuration, called sesquialtera. This is represented in the data
with the same rhythmic duration in each part, usually as a whole
note and a triplet dotted whole note using an *rscale:3/2
interpretation in the second part to indicate that it is to be
displayed with a rhythmic duration 3/2 longer (i.e., convert the
triplet dotted whole note into a non-triplet dotted whole note).
In mensural music (particularly in masses) the voices are sometimes
notated in conflicting note values, related in part to using less
common mensurals signs such as C-dot and O-dot. (Colloquially we
refer to a voice being "augmented" or "diminished" relative to the
others.) The most common situation is for a half-note in the tenor
to have the same duration as a whole-note in another part. The
*rscale
rhythmic-scaling interpretation will also be used in these
cases. All parts in the data have a 1:1 rhythmic scaling in the
modern sense against each other for analytic purposes. Applying
the *rscale
factor will cause a part to be displayed in its
original rhythmic values (i.e., reverse the resolution of the
augmentation/diminution).
Triplet brackets are given explicitly in the data using "V" and "Z" characters. Triplet brackets indicate music that was originally in colored notation, which is a mensural equivalent of triplets (although coloration can be used for other purposes as well). Music in 3-mensuration (an alternate mensural method of showing triplets) may be represented as modern rhythmic triplets in the data, but will not contain triplet brackets. Mostly commonly 3-mensurtion music is represented in 3/1 modern time signatures to be played 2/3rds faster, while colored notation are typically represented as triplets in 2/1 time signature. When the perceived beat is at the triplet wholenote level in colored notation, the modern time signature is given as 3/3%2 (which is not a standard time signature in modern notation). 3%2 represents 2/3rds of a whole notes, or in other words a triplet whole note. So a 2/1 measure has the same duation as 3/3%2 (two wholenotes equals three triplet wholenotes).
Ligatures are groups of notes with their heads attached to each other which typically gives information about the rhythm of the notes in mensural notation. Usually modern notation of mensural music will place brackets around individual notes which were originally part of a ligature group. However, ligatures are typically not indicated in this data.
Individual composer's works are archived independently, and each of their repositories can be viewed by clicking in the first column of the table below. Each composer is assigned a unique three-letter code listed in the first column that is the name of the directory in the meta-repository which contains each composer's works.
Ano | Anonymous | (JRP) | (Wikipedia) |
Agr | Alexander Agricola | (JRP) | (Wikipedia) |
Bru | Antoine Brumel | (JRP) | (Wikipedia) |
Bus | Antoine Busnoys | (JRP) | (Wikipedia) |
Com | Loyset Compere | (JRP) | (Wikipedia) |
Das | Ludwig Daser | (JRP) | (Wikipedia) |
Duf | Guillaume Du Fay | (JRP) | (Wikipedia) |
Fry | Walter Frye | (JRP) | (Wikipedia) |
Fva | Antoine de Fevin | (JRP) | (Wikipedia) |
Gas | Gaspar van Weerbeke | (JRP) | (Wikipedia) |
Isa | Henrich Isaac | (JRP) | (Wikipedia) |
Jap | Jean Japart | (JRP) | (Wikipedia) |
Jos | Josquin des Prez | (JRP) | (Wikipedia) |
Mar | Johannes Martini | (JRP) | (Wikipedia) |
Mou | Jean Mouton | (JRP) | (Wikipedia) |
Obr | Jacob Obrecht | (JRP) | (Wikipedia) |
Oke | Johannes Okeghem | (JRP) | (Wikipedia) |
Ort | Marbrianus de Orto | (JRP) | (Wikipedia) |
Pip | Matthaeus Pipelare | (JRP) | (Wikipedia) |
Reg | Johannes Regis | (JRP) | (Wikipedia) |
Rue | Pierre de la Rue | (JRP) | (Wikipedia) |
Tin | Johannes Tinctoris | (JRP) | (Wikipedia) |
Within each directory of the repository is a subdirectory called
kern
that contains the actual scores in the **kern
data type
of the Humdrum data file format. MIDI files can be downloaded from
the JRP website by typing the command make midi
in the root
directory of the jrp-scores repository (provided that you have GNU
make installed on your computer). PDF files for the musical scores
(which are generated from the Humdrum data files found in this
repository) can be downloaded with the command make pdf
. PDF
scores are printed with the MuseData
printing system. A command-line version of this system is available
at http://muse2ps.ccarh.org . MuseData generated from the Humdrum
data files can be downloaded with the command make musedata
.
Currently, composers with complete (or nearly complete) sets of encoded works include: Gaspar, Josquin, La Rue, Okeghem, and de Orto.
The relative size of each composer's work database can be viewed on the JRP Census page. You can count the number of works and notes for each composer by the command:
make note-counts-by-composer
Each file in the database starts with a unique JRP catalog number.
This consists of a three-letter composer ID as listed above, followed
by a four-digit number to represent a specific work by that composer.
Typically the first digit of the catalog number will indicate the
genre of the work: 1 for masses, 2 for motets, and 3 for secular
songs (see the !!!AGN
reference record inside of the file for the
actual genre designations). The music of Josquin is an exception,
where the first two digits indicate its volume number in the New
Josquin Edition, and the last two digits indicate the enumerated
position of the work within that volume.
After the four-digit work number, an optional letter indicates that a movement-level encoding of a work is found in the file. This system is used for storing mass sections in separate files, where "a" is the first section (usually the Kyrie section), "b" is the second section (usually the Gloria section), and so on.
Work-level variants are indicated in the catalog number by adding a dot after the work number, followed by a variant number (of any digit length). Variants at the movement level are indicated by placing a dot after the variant letter(s) followed by a variant letter (or more than one lower-case letter). If a file contains an enumerable subsection of a movement, an optional digit will be appended to the catalog number (which may itself be followed by a dot and variant digits). Movement subsection numbers are not yet present in any datafile (and neither are Movement sub-subsections which would alternate back to letters like movement-level labels).
After the catalog number, each filename contains a dash followed
by title information. The title information may have up to three
fixed fields, each separated by a dash from the other. The first
component of the title information is the title of the work. This
is optionally followed by the title of the movement (i.e., mass
section names, such as the standard sections: kyrie, gloria, credo,
sanctus, and agnus [dei]). A third optional section of the title
in the filename is variant information. Spaces in the title are
encoded as underscore characters in the filenames. Two underscores
in a row indicate a slash in the title. No accent marks are given
in the filename titles since they are
POSIX compliant; see the
!!!OTL
reference record within each file for the proper accentuation
of work titles.
Title information in filenames are a courtesy for human beings. As such they can be removed from filenames, leaving only the unique catalog numbers. (Removing catalog numbers will not result in unique filenames since a composer may write several songs with the same title).
To download this Github repository using git in a terminal, type:
git clone --recursive https://github.com/josquin-research-project/jrp-scores
The --recursive
option is needed to download each of the
individual composer repositories inside of the meta-repository.
In a unix terminal, you can check to see if git is installed by
typing which git
. If the terminal replies with a path to
git, then you can proceed with the above cloning to download the
repository. If not, then typically you can use a package manager
to install git, such as apt-get install git
or yum install git
in linux. On Apple OS X computers, git can be installed
directly from here or by more
experienced users from a mac package manager such as
Homebrew. If you have a comicbook-like view of the
computer world, you can download GUI interfaces for git
here. A Github/git
plugin is also available for the Eclipse
IDE (watch video).
This repository collects individual composer repositories as submodules. You can also download the composer repositories separatly, such as for Josquin:
git clone http://github.com/josquin-research-project/Jos
After you have downloaded this repository with git
, you can check
periodically for updates for all composers' works using this command
in the jrp-scores
directory:
make update
The digital scores in this repository are designed to work with the Humdrum Toolkit (github). A makefile in the base directory of the repository contains some basic processing commands which either require Humdrum Extras (github) to manipulate the data files, or commands (starting with "web") which download data generated online by the JRP website.
Here are some of the make commands which you can run in the base directory of the downloaded repository:
No additional software needed: | |
make | List all of the possible make commands (i.e., this list). |
make update | Download any updates to the online repository. |
make clean | Delete directories of data created by this makefile, such as kern-reduced, midi, pdf, pdf-notext. |
make notitle | Remove titles from files names and store in directory called kern-notitle in each composer's directory. |
make web-pdf | Download PDF files for graphical music scores for each piece from the JRP website. |
make web-pdf-notext | Download PDF files for graphical music scores for each piece from the JRP website with lyrics removed from all parts. |
make web-reduced | Download version of the data file which divides all note durations by a factor of four. This data is useful for doing rhythmic analysis with the standard Humdrum Toolkit. |
Humdrum Extras installation required: | |
make reduced | Decrease all note durations by a factor of four. Output data will be stored in a directory called kern-reduced within each composer's directory. Similar to make web-reduced, but much faster. |
make notext | Remove lyrics from all parts. Resulting data will be stored in a directory called kern-notext within each composer's directory. |
make genres | Download works organized by genre from kernScores. |
Humdrum Toolkit installation required: | |
make census | Run census -k on all works. |
For proper rhythmic parsing in the Humdrum Toolkit, some files containing rational rhythmic values need to be diminuted by a factor of four to convert whole notes (typically the rhythmic level of a beat in the early Renaissance), into quarter notes (the typical modern rhythmic level for beats). This can be done with the rscale tool to apply a rhythmic scaling of 1/4 to all notes in all scores:
rscale -f 1/4 oldfile > newfile
A makefile in the base directory of this repository (this one contains instructions to generate reduced rhythmic versions of all files. Type (if Humdrum Extras is installed):
make reduced
or if humextras is not installed, type:
make web-reduced
Here are some other interesting processing actions:
- Determine works that include text/lyrics:
grep -rl '\\\*\\\*text' \*/kern/\*
The website http://josquin.stanford.edu is a high-level interface to these scores, but also includes lower-level access to the data, data conversions, and analytic tools using URL parameters from the https://data.josqu.in web address. On the JRP website, each score has a "work info" page generated with this format:
https://data.josqu.in/work?id=Jos2721.krn
Example MIDI data file access by setting the file extensiom to mid
:
http://data.josqu.in/Jos2721-La_Bernardina.mid
Example graphical music scores in PDF format with four possible styles:
- Without editorial accidentals or text: http://josquin.stanford.edu/data?a=notationnoedit&f=Duf1005a-Missa_Se_la_face_ay_pale-Kyrie
- Without editorial accidentals but with text: http://josquin.stanford.edu/data?a=notationnoeditwithtext&f=Duf1005a-Missa_Se_la_face_ay_pale-Kyrie
- With editorial accidentals but without text: http://josquin.stanford.edu/data?a=notationwitheditorial&f=Duf1005a-Missa_Se_la_face_ay_pale-Kyrie
- With editorial accidentals and with text: http://josquin.stanford.edu/data?a=notationwitheditorialwithtext&f=Duf1005a-Missa_Se_la_face_ay_pale-Kyrie
Scores are generated from the original Humdrum **kern
scores found
in this repository for each work, so music notation in the PDF files
should match exactly. Online PDF files of the scores are created
using the hum2muse converter
which translates Humdrum files into MuseData file which can then
be converted into PostScript files with
muse2ps.
The kernScores library of musical scores for analysis in the Humdrum Toolkit has a page dedicated to the JRP scores:
http://kern.humdrum.org/browse?l=jrp
The Humdrum Extras tools/library
(github) contains internal
knowledge for how to download the digital scores on the command
line. To download from the JRP website, the filename is prefixed
by jrp://
. To download from the kernScores website, the
prefix is humdrum://
, or h://
for short. KernScores
access requires the exact file name (catalog number, title, file
extension), while JRP access requires only the catalog number.
Examples:
humcat h://jrp/Jos/Jos2721-La_Bernardina.krn
humcat jrp://Jos2721-La_Bernardina.krn
humcat jrp://Jos2721
Since it requires the full filename for accessing individual files, the kernScores downloading method is instead mostly useful for downloading an entire set of composer's works. Try the following humextra command to download all of the works for Okeghem:
mkdir Oke; cd Oke; humsplit h://jrp/Oke
The kernScores website has five meta-collections of the scores: two for Josquin authenticity levels, and three for the main genres of mass, motet and song. These can be downloaded like this on the command line:
Secure Josquin set: |
mkdir -p Joa/kern; (cd Joa/kern; humsplit h://jrp/Joa)
|
Not secure Josquin set: |
mkdir -p Job/kern; (cd Job/kern; humsplit h://jrp/Job)
|
All mass sections: |
mkdir -p Zma/kern; (cd Zma/kern; humsplit h://jrp/Zma)
|
All motets: |
mkdir -p Zmo/kern; (cd Zmo/kern; humsplit h://jrp/Zmo)
|
All songs: |
mkdir -p Zso/kern; (cd Zso/kern; humsplit h://jrp/Zso)
|
Type make genres
in the base directory of the repository if
Humdrum Extras is installed to
download the Zma, Zmo, and Zso genre groupings from the kernScores website.
Type make Joa; make Job
in the base directory of the repository if
Humdrum Extras is installed to
download the secure and not secure groupings of Josquin music.