Audio-to-midi

An application of melody extraction.

Core algoithm

(MSnet): https://github.com/bill317996/Melody-extraction-with-melodic-segnet

Dependencies

Requires following packages:

• python 3.6
• pytorch 1.0.0
• numpy
• pysoundfile
• scipy
• math
• pypianoroll

Usage

Put your audio files in "./input/" folder and run

python audio2midi.py

or

python3 audio2midi.py

Results will save in './output/'

audio2midi.py

usage: audio2midi.py [-h] [-in INPUT_FOLDER] [-out OUTPUT_FOLDER] [-m]

Audio to midi : Update in 20190503

optional arguments:
  -h, --help          show this help message and exit
  -in INPUT_FOLDER    path to input folder (default: ./input/)
  -out OUTPUT_FOLDER  Path to output folder (default: ./output/)
  -m                  Extract main-melody instead vocal-melody (default: False)