Due to the widespread noise pollution in our cities, Sunbird AI recognised a need to gather data on noise exposure in order to produce an action plan and empower citizens to be vigilant in tracking and monitoring noise.

This process was carried out by collecting noise samples around Kampala using mobile phones with the Open Data Kit (ODK) data collection tool. Sunbird AI gathered a total of nineteen noise data categories collected from the six divisions of Kampala, and for each of the nineteen categories we collected an average of 2000 noise data points.

The challenge is to train a noise classification model to classify noise into noise categories

There are ~ 30 000 audio files in train and ~12 000 audio files in test.