Instruments Activity Detection using Spleeter and Auditok

Detect individual instruments activity in an audio file.

This project relies on spleeter for performing source separation and on auditok for the activity detection.

Installation

for Auditok run

git clone https://github.com/amsehili/auditok.git
cd auditok
python setup.py install

then run

pip install -r requirements.txt

Usage

Provide an audio file

python run.py -a my_audio_file.mp3

use youtube-dl to download from a query of [artist] [song]

python run.py david_bowie life_on_mars

Output

start-end-duration of each instrument is saved as [instrument].txt in the data dir.

Instruments Timeline

     start tags
0      8.0   🎤🎶
265   10.0    🎶
140   10.0    🎹
266   12.0    🎶
210   12.0    🎸
..     ...  ...
136  239.0    🥁
137  239.0    🥁
264  239.0    🎸
138  241.0    🥁
139  243.0    🥁

 drums 

   start    end  duration
0  35.20  35.51      0.31
1  41.97  42.97      1.00
2  46.13  46.48      0.35
3  49.47  49.83      0.36
4  56.91  61.20      4.29
-------------------

 piano 

   start    end  duration
0   9.99  10.33      0.34
1  11.96  16.96      5.00
2  16.96  21.96      5.00
3  21.96  26.96      5.00
4  26.96  29.23      2.27
-------------------

 vocals 

   start    end  duration
0   8.21  12.98      4.77
1  14.47  17.52      3.05
2  18.61  21.50      2.89
3  22.75  25.95      3.20
4  26.96  30.21      3.25
-------------------

 other 

   start    end  duration
0   8.21   9.21      1.00
1   9.68  11.09      1.41
2  11.87  13.16      1.29
3  13.24  15.55      2.31
4  15.58  20.58      5.00
-------------------

 bass 

   start    end  duration
0  11.89  14.92      3.03
1  14.94  19.55      4.61
2  21.23  23.37      2.14
3  31.82  34.74      2.92
4  35.68  36.83      1.15
-------------------

Instruments and their durations:

other 231.89
bass 206.09
drums 161.37
vocals 159.87
piano 107.38