An automatic annotation tool of piano fingering based on HMM.
Example command line:
python autofingering sheet/sample.mxl -ddata -oparams
usage: autofingering [-h] [-o OUT_FILE] [-i IN_PARAMS] [-d DATA] [-p OUT_PARAMS]
[--right-beam RIGHT_BEAM] [--left-beam LEFT_BEAM]
input_sheet
Automatic annotation of piano fingering based on HMM.
positional arguments:
input_sheet Input music XML file name.
optional arguments:
-h, --help show this help message and exit
-o OUT_FILE, --out-file OUT_FILE
Annotated output music XML file name.
-i IN_PARAMS, --in-params IN_PARAMS
File name of pre-trained hand model params.
-d DATA, --data DATA PIG data directory path.
-p OUT_PARAMS, --out-params OUT_PARAMS
File name of model parameters. It is only necessary to save the learned
parameters.
--right-beam RIGHT_BEAM
Specify right hand beam number.
--left-beam LEFT_BEAM
Specify left hand beam number.
- pandas
- numpy
- music21
PIG format files with finger numbers are valid. You can download sample data from here.
They should be written in MusicXML format.
The first page of Chopin etude Op.10 No.4 in C Minor "Torrent".
All finger numbers are auto-generated.
Nakamura, E., Saito, Y., & Yoshii, K. (2020). Statistical learning and estimation of piano fingering. Information Sciences, 517, 68-85. doi:10.1016/j.ins.2019.12.068