This tool allows you to visualize and compare some audio classification models. The screenshots below show how the audio player works.
To customize it you have to modify only the config.json file. The directory structure to comply with is the following:
-| annotation_dir
-| model_name_1
-| song_1.json
-| song2.json
-| ...
-| model_name_2
-| song_1.json
-| song2.json
-| ...
-| ...
Each .json file have comply with the following structure:
[
{
"start": 0,
"end": 24.79891156462585,
"data": {
"label": "instrumental"
}
},
{
"start": 24.822131519274375,
"end": 29.326802721088434,
"data": {
"label": "vocals"
}
},
...
]
The config.json file contains the following parameters:
{
"app_name": name of the system (ex. Instrumentalness),
"song_names": an array of the song names available, without the file extension,
"annotation_dir": absolute path of annotation_dir folder,
"audio_dir": absolute path of the folder with the audio file to listen to
}
To run this application you have to:
- Clone this repository
- cd AudioPlayer
- npm install
- npm start