search

JuanFMontesinos / Acappella-YNet

Official implementation of A cappella: Audio-visual Singing VoiceSeparation, from BMVC21
https://ipcv.github.io/Acappella/
14 stars 4 forks source link

There is no dataset config ? #3

Closed Enescigdem closed 2 years ago

Enescigdem commented 2 years ago

Hi, In README. md , Download the code and set your dataset paths at config>dataset_paths.json is said but there is no such file in code directory or it is not mentioned about the format of that json file. Thanks in advance

EmreOzkose commented 2 years ago

They are added this file to gitignore file. Hence this file is not uploaded. The reason is, I think, private path issue. You can create a file configs/

where 

{
    "acappella": "/path/to/data/acapella_preprocessed/splits",
    "audioset": "/path/to//data/audio"
}

the folder splits is already created by preproc/preprocess.py. the folder audio contains folder eval. folder eval contains folder audio. folder audio contains some audios which are acapella. These folders are download in https://research.google.com/audioset/download.html. You should download 3 file:

eval_segments.csv Balanced train - balanced_train_segments.csv Unbalanced train - unbalanced_train_segments.csv

and locate to preproc.

Then you should run this :

streamlit run audioset.py

Then you should copy downloaded files to /path/to//data/audio/eval/audio.

Let me know if there is a wrong step @JuanFMontesinos .

JuanFMontesinos commented 2 years ago

Indeed I didn't realise the file was in gitignote and the structure should be like that. Should be fixed at 1dbe364