Example of extraction configuration file

The following code shows the available configuration and all the default values. Remember that you can always add your own variables and you will find them in the ExtractConfiguration object!

# Directory containing MusicXML files to be analized. By default, the extension should
# be .xml, but you can change it via `musif.musicxml.constants.MUSICXML_FILE_EXTENSION`
# Note that MusicXML files supported are limited to Finale exported ones.
data_dir: null

# Directory containing MuseScore files to be analyzed. By default, the extension should
# be .xml, but you can change it via `musif.musescore.constants.MUSESCORE_FILE_EXTENSION`
# If only musescore_dir is give, musif will try to convert the files to MusicXML using
# musescore -- see `mscore_exe` setting.
# If xml_dr is given, musescore files are only used if one of the following features are
# requested: [ harmonic, scale_relative ]. In this case, only the harmonic annotations
# are ridden from MuseScore files.
musescore_dir: null
  
# Path to the binary of musescore. Used in case only MuseScore files are provided to
# convert them to MusicXML. Note that most of the features won't work with those files.
# Prefer using Finale MusicXML files, instead.
# This option can be:
# a) null -> ms3 library will try to find the musescore binary by itself
# c) string -> the path to the mscore executable
# b) list or tuple -> each element is a string as in `subprocess.run()` use it for remote
#    server processing such as `xvfb-run -a mscore`
mscore_exec: null

# Directory to keep .pkl files if caching is enabled. null for disabling cache.
# See `Caching` page for more info
cache_dir: null
  
# If not null, musif will export MusicXML files to dataframes pickled in this directory.
# See `musif.cache.utils.store_score_df` for more info.
# They are useful for allowing interoperability with data science packages.
dfs_dir: null

# Boolean flag to expand repetitions of the music score or not.
expand_repeats: false

# Size of windows (in number of measures) when extracting features of the score in a.
# Each row of the resulting DataFrame will correspond to each window.
# If null, only one window is extracted, with the  size of the whole score.
# A `WindowId` column is always added, containing the number of the window in the score
# to which a row of the dataframe refers to.
# Note that the standard `Id` column referred to the MusicXML files is always present.
window_size: null

# Size of overlap (in measures) beatween each window. Considered only if `window_size` is
# not null. It must be less than window_size - 1!
overlap: 2

# Number of parallel processes to be used as defined by joblib: 
# 1 => no parallel. 2 => 2 processes. -1 => all available virtual cores. -2 => all
# available virtual cores except 1
parallel: 1

# Directory to save the logs files, level for logging in the console (console_level) and
# inthe log file (file_level)
log:
  file_path: ./musif.log
  file_level: DEBUG
  console_level: ERROR

# Modules that are run once for every score. This distinction is important only when
# `window_size` > 1
# They are processed first and used by the other features so disabling them might result
# in errors when extracting features.
# Dependencies may be expressed in the `musif_dependencies` property of the feature, see
# for instance `musif.extract.features.density.__init__.py`
basic_modules: ['scoring']
# Other values: file_name_generic

# Modules to be extracting from scores. Order is important, so last ones will be
# processed last. Some of them might depends on others.
# For example, density model relies on the number of note, which is computed in core.
# Dependencies may be expressed in the `musif_dependencies` property of the feature, see
# for instance `musif.extract.features.density.__init__.py`
# To disable/enable a module, simply comment the line.
features: ['core']
  # Possible values:
  # - core
  # - ambitus
  # - melody
  # - tempo
  # - density
  # - texture
  # - harmony # require MuseScore files with harmonic annotations
  # - lyrics
  # - scale
  # - scale_relative # require MuseScore files with harmonic annotations
  # - key
  # - dynamics    
  # - rhythm
  # - music21
  # - jsymbolic

# The list of packages/modules where basic features are. See page `Custom Features` for
# more info
basic_modules_addresses: ["musif.extract.basic_modules"]

# The list of packages/modules where features are. See page `Custom Features` for
# more info
feature_modules_addresses: ["musif.extract.features"]

# List of hooks executed before of creating the read-only cache objects.
# See `Caching` page for more info
precache_hooks: []

# remove notes and objects referred to percussion instruments
# (musif stock features do not work with percussion instruments)
remove_unpitched_objects: true

# Filter to select only some instruments to be processed for each score. 
# If the filter is None (disabling all parts), the program will extract all parts.
parts_filter: []
  # Possible Values. If non it selected, all parts will be computed.
  # - vnI
  # - vnII
  # - obI
  # - obII
  # - va
  # - bs
  # - sop
  # - ten
  # - alt
  # - bar
  # - bass
  # - bbar

# Used to split the possible layers present depending on the instrument name
split_keywords: []
  # Possible Values
  # - woodwind
  # - brass
  # - wind
  # - viola
  # - cello
  # - bass
  # - oboi