data-music

This repository contains the MEI-encodings of the music sources of the Freischütz Digital project.

Introduction

In order to handle the desired complexity of information, Freischütz Digital came up with an MEI-based data setup called Core model. The intention is to separate substantial musical information from the mere graphic appearance of the music manuscripts and sources – if one source uses some kind of visual shortcut to abbreviate the very same musical content that another source has written out, there is no substantial difference between these sources. Accordingly, our encodings should not be bloated by capturing graphical details of limited musical meaning. As MEI uses TEI's concept of parallel segmentation for encoding variation between multiple sources, the use of <app> and <rdg> can make an encoding unmanageable quite easily. At the same time, Freischütz Digital is specifically interested in preserving information about the scripture of the sources. This results in a conflict of interests. One extreme would be to encode all differences using <app> and <rdg> in one single file, no matter how hard it would be to maintain and process this file. The other extreme would be to encode all musical sources separately, and then use external pointers etc. to identify relevant similarities and differences from the outside. It should be clear that this setup becomes equally complex when used to store the same amount of information. Freischütz Digital seeks a compromise between both approaches, effectively separating substantial musical content from it's graphical presentation in the manuscripts and prints. Admittedly, this model is also very complex, but ultimately, the complexity of a data model will always have to match the complexity of the contents that are to be encoded by this model.

The Core Model

The basic idea of Freischütz Digital's data model is to put the musical content of all considered sources into one shared file, which serves as core of the whole model. If sources differ in pitches, durations, slurs and similar substantial information, this variation is encoded in the core using parallel segmentation with <app> and <rdg>. This file does not contain any layout-related details like information about abbreviations and scribal shortcuts ("colla parte" etc.) or stem directions. Besides the core, there is an additional file for each source which duplicates the complete document tree from the core, this time only containing the variants relevant to this particular source (that is, there are no <app> and <rdg> elements in these files, but only the reading from the respective source). Each <note> (and all other musical events as well) has no substantial information, so there are no @dur, @pname, @oct etc. attributes. Instead, these elements just provide graphical information (like @stem.dir on <note> or @curvedir on <slur>). Additionally, every single element points to its counterpart in the core file using a @sameas relation. In order to fully understand an element in the sources, a user has to follow its @sameas pointer to the core and copy in all other attributes provided there. This means you can look at the musical content by processing just the core, but you need to process both the core and the souce file to get the full encoding of a source. For convenience, this repository contains some pre-processed files, which may be easier to use in most cases. However, if you want to change some data, you have to do your changes to the underlying files, and just regenerate the usable files from there. The following section explains all that.

Repository Structure

First, the core file is called freidi-work.xml and lives at the top level of this repo. All sources are contained in the musicSources folder, and have names like freidi-musicSource_A.xml. The last part of the file name matches the siglum of that source, which is also the @xml:id of the <mei> of each file. Usually, you have to modify only these files when changing data, while all other files in the repo should stay unchanged. However, it is extremely important to make sure that the information in the core and the sources match!!! But no worries, validation tries to help you as much as possible. Just make sure your changes still validate against both the RelaxNG and Schematron rules, and you're almost good to go. Here's a complete list of files and folders in the repo:

• freidi-work.xml: This is the main file resembling the core of our data model. It contains the musical content, but also the most detailed information about the relation of sources etc., plus editorial annotations. Changes to the Freischütz data normally require to edit this file.
• musicSources/: This folder holds the sources of our data model. Each source has a separate file in here. Changes to a source require to edit the corresponding file in this folder.
• schemata/: This folder holds the ODD source code for the schema we use: freidi-music-source.rng. This file also contains a whole bunch of Schematron rules, which are essential for validation.
• xslt/: This folder contains four XSLT stylesheet which can be used to pre-process our data from the core model to something more common for other use cases. Actually, our edition at http://freischuetz-digital.de/edition/ depends on these pre-processed files. The XSLTs are: -- buildAnnotations.xsl: This stylesheet is operated on freidi-work.xml. It extracts all editoral annotations into a format compatible with Edirom Online. -- buildEdiromMeasures.xsl: This stylesheet operates on the source files. It throws away all musical content inside <measure>, but keeps the measure positions on the facsimile pages. The resulting files are used to browse the sources, jump to individual measures etc. -- buildMeasureConcordance.xsl: This file operates on freidi-work.xml and looks for every <measure>. It then processes all sources and identifies which measures refer to the one in the core. Then, a list of all these relations is generated. -- buildMusic.xsl: This stylesheet operates on the source files. In essence, it compiles the source with the content from the core. It will output multiple files of different flavors for each movement of that source that actually contains music. First, it creates files in source_abbr, where all abbreviations and shortcuts of the sources are kept. This partners with source_expan, where all abbreviations will be resolved. Both forms are combined in source_raw, which still contains <choice> and similar elements. If the stylesheet is used on freidi-musicSource_A.xml, there will be additional files which split up the core by individual movements. All XSLTs output there results to a folder called ready-for-use.
• ready-for-use/: This folder contains various files which may be more appropriate for direct use. These files are described in the section above. However, it is important to not modify these files directly, but instead change the core and sources directly, and then use the provided XSLTs to re-generate them.

The state of this data

The files in this repo reflect the current state of the Freischütz Digital data. Even though the project has officially ended, they are still subject to changes. Major revisions will be made available through TextGrid for increased sustainability, while "daily" work will happen here. The encoded sources in this repo have a different "level of completeness".

• freidi-musicSource_A.xml: Complete encoding of the whole opera.
• freidi-musicSource_KA1.xml: Complete for movements 6, 8 and 9, measure positions for the others.
• freidi-musicSource_KA2.xml: Complete for movements 6, 8 and 9, measure positions for the others
• freidi-musicSource_KA9.xml: Complete for movements 6, 8 and 9, measure positions for the others
• freidi-musicSource_K13.xml: Complete for movements 6, 8 and 9, measure positions for the others
• freidi-musicSource_K15.xml: Complete for movements 6, 8 and 9, measure positions for the others
• freidi-musicSource_KA19.xml: Complete for movements 6, 8 and 9, measure positions for the others
• freidi-musicSource_K20.xml: Complete for movements 6, 8 and 9, measure positions for the others
• freidi-musicSource_KA26.xml: Complete for movements 6, 8 and 9, measure positions for the others
• freidi-musicSource_D1849.xml: Complete for movements 6, 8 and 9, measure positions for the others
• freidi-musicSource_ED-kl.xml: Measure positions only.
• freidi-musicSource_D1849B.xml: Measure positions only.
• freidi-musicSource_D1867.xml: Measure positions only.
• freidi-musicSource_D1871.xml: Measure positions only.
• freidi-musicSource_DJochum.xml: Measure positions only.

License

All MEI-files in this repository are available under a Creative Commons CC-BY-NC-SA license. The XSLTs are available with a AGPLv3 license. However, if you would like to use our encodings or stylesheets for something that isn't possible with these license, please contact us.