The Filk Archive

This directory contains a number of utilities and notes for use by the Filk Archive project.

The program zoom-mixdown.sh mixes down Zoom H2 and H2n 4-channel recordings to 2-channel stereo.

The program split-audio.pl splits and encodes an audio file (in any format readable by SoX) into mp3 clips, suitable for zooniverse, and also outputs a csv file formatted appropriately for the zooniverse panoptes tool. You must have the SoX toolchain installed, as well as the Perl DateTime module. (On Cygwin, both of these can be installed as Cygwin packages with the Cygwin setup tool.)

The program analyze-zooniverse.pl processes the output of the zooniverse classifications.csv into a csv file suitable for human consumption (when imported into a spreadsheet).

To generate sha256 checksums of all the files in a directory:

$ find . -name sha256-sum.txt -o -type f -print0 | xargs -0 sha256sum >> sha256-sum.txt

To create flac files from all the wav files in a directory:

$ find . -iname '*.wav' -print0 | xargs -0 flac -V --keep-foreign-metadata -f --delete-input-file

Uploading split files to panoptes

To split files, run the split-audio file something like:

$ ~/split-audio.pl --recorded-by="J. Spencer Love" --metadata="British Filk Convention: Obliter-8"/metadata.json --output-dir=/cygdrive/g/Split/Obliter-8 "British Filk Convention: Obliter-8"/*.wav

If you want to host the split files somewhere other than on Zooniverse itself, also add an option like

--output-url=https://www.ladymondegreen.com/filkarchive/Obliter-8/

The metadata file is a JSON file containing metadata both for the entire "location" (Convention), and the individual "events" (recordings). For example the metadata for the above invocaton looks like:

{
    "location": "British Filk Convention: Obliter-8",
    "date": "31 Jan - 2 Feb 1997",
    
    "British Filk Convention: Obliter-8/Obliter-8 #1.wav": {"event": "Tape 1"},
    "British Filk Convention: Obliter-8/Obliter-8 #2.wav": {"event": "Tape 2"},
    "British Filk Convention: Obliter-8/Obliter-8 #3.wav": {"event": "Tape 3"},
    "British Filk Convention: Obliter-8/Obliter-8 #4.wav": {"event": "Tape 4"},
    "British Filk Convention: Obliter-8/Obliter-8 #5.wav": {"event": "Tape 5"},
    "British Filk Convention: Obliter-8/Obliter-8 #6.wav": {"event": "Tape 6"},
    "British Filk Convention: Obliter-8/Obliter-8 #8.wav": {"event": "Tape 8"}
}

Once the files are produced, they need to be uploaded to Zooniverse using the panoptes tool. Change to the output directory specified above, and then run (e.g.)

$ panoptes subject-set upload-subjects 78566 subjects.csv

The correct subject-set ID must be chosen; subject-sets can be listed with

$ panoptes subject-set ls -p 9901

The panoptes tool can be installed using Python's pip utility, as

$ pip install panoptescli

If the split files are being hosted somewhere other than Zooniverse (i.e. the --output-url option was used when generating the split files), first upload the split files to the hosting server, e.g.

$ rsync -avP *.mp3 [email protected]:ladymondegreen.com/filkarchive/Obliter-8

Information about them can be uploaded to Zooniverse using the panoptes tool.

$ panoptes subject-set upload-subjects -r 2 -m audio/mpeg 78566 subjects.csv