Urban sound quality recognition using smartphones
Master Thesis
Author
Στεφόπουλος, Ανδρέας
Stefopoulos, Andreas
Date
2023-02View/ Open
Abstract
Urban sound quality is a topic that affects the life and well-being of all living creatures, including us humans. Noise pollution in cities is becoming more and more present, resulting in low sound quality levels. This research project aims to make this phenomenon more visible using artificial intelligence tools and provide insights into how humans' perceptions is functioning under a specific set of sounds. For this purpose, data were collected using a smartphone application and annotated using the perceived sound quality of the user in different locations. Furthermore, experiments were conducted using audio analysis and neural networks to create a model that can classify the sound quality based on the perceived qualities and provide results and insights based on the user’s location. Lastly, as a final result, an end-to-end process has been made, which could potentially be used every day in different locations, gathering sounds and classifying the quality, aiming to track the change of the quality also in the course of time.