I'm new to Python, but I've been able to easily separate instruments and vocals using Python tools, so I'll share them.
Install Python by referring to the following article
$ brew install pyenv
$ echo 'eval "$(pyenv init -)"' >> ~/.bash_profile
$ exec $SHELL -l
Since the project uses Python3, install anaconda3
$ pyenv install anaconda3-2019.10
$ pyenv global anaconda3-2019.10
https://qiita.com/aical/items/2d066801a7464a676994
The tool used this time is spleeter https://github.com/deezer/spleeter Reference video https://www.youtube.com/watch?v=HqanOjPBRAQ
$ git clone https://github.com/Deezer/spleeter
$ conda install -c conda-forge spleeter
This completes the preparation.
Put the sound source you want to divide in the spleeter folder of the project file.
$ spleeter separate -i spleeter/(File name of the sound source you want to divide).mp3 -p spleeter:2stems -o output
When this command is executed and the process is completed, an output folder will be created in the same directory as spleetr. If the instrument and vocal files are included in it, it is successful.
Male vocals were a little difficult to divide. The female vocal was missing better. According to a friend, before separating the instrument and vocals with this software, it seems easier to remove the bass by using the equalizer in the composition software.
Recommended Posts