Skip to content


Switch branches/tags

Name already in use

A tag already exists with the provided branch name. Many Git commands accept both tag and branch names, so creating this branch may cause unexpected behavior. Are you sure you want to create this branch?

Latest commit


Git stats


Failed to load latest commit information.
Latest commit message
Commit time

JRP Digital Scores

This repository contains digital music scores in the Humdrum data format from the Josquin Research Project. The primary web interface for these scores is 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 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 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 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.

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.

All parts are encoded in modern notation that is 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. 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. (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 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. 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, 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. 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 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)
Ock Johannes Ockeghem (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 . 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: Josquin, Ockeghem, La Rue and de Orto.

The relative size of each composer's work database can be viewed on the JRP Census page.


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

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 cannot be downloaded in a very useful format from the ZIP link on the Github website, since the included repositories for each composer will not be included in that ZIP file. You must either use git software or separately download each of the individual composer's ZIP files linked from the table above. Use this bash script to download manually using wget (usually for linux & cygwin), or this bash script for OS X.


After you have downloaded this repository with git, you can check periodically for updates for all composers' works using this command:

git pull --recurse-submodules
git submodule foreach git checkout master

Alternatively, the makefile in the base directory can be used to run this command:

make update

If the scores were downloaded manually with one of the two gist scripts below, you must re-download all files to check for updates. Re-run one of these two scripts using either wget (linux & cygwin) or curl (Apple OS X).

Processing scores

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.

Rhythm representation considerations

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

to download the reduced form from the JRP website (example). This will create subdirectories in each composer directory named kern-reduced with the reduced rhythmic versions of the data files.

Other example processing

Here are some other interesting processing actions:

  • Determine works that include text/lyrics: grep -rl '\*\*text' */kern/*

Alternate data access

JRP website

The website 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 in the web address. On the JRP website, each score has a "work info" page generated with this format:

Where Jos2721 is the JRP catalog number 2721 from Josquin's collection of works, and info is the action being applied to the given work (file) which causes the work-info page to be displayed. A title may optionally follow the catalog number (with exact spelling of the title in the database):

A file extension can also be given (.krn in this case) but will always be ignored. Example Humdrum data file access by setting the action to "humdrum":

Example MIDI data file access by setting the action to "midi":

Example graphical music scores in PDF format with four possible styles:

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.

kernScores website

The kernScores library of musical scores for analysis in the Humdrum Toolkit has a page dedicated to the JRP scores:

Humdrum Extras tools/library

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.


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 Ockeghem:

mkdir Ock; cd Ock; humsplit h://jrp/Ock

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.


Digital scores for all composers in the Josquin Research Project.







No releases published


No packages published