Title: Automatic Organisation and Quality Analysis of User-Generated Content with Audio Fingerprinting

Abstract: The increase of the quantity of user-generated content experienced in social media has boosted the importance of analysing and organising the content by its quality. Here, we propose a method that uses audio fingerprinting to organise and infer the quality of user-generated audio content. The proposed method detects the overlapping segments between different audio clips to organise and cluster the data according to events, and to infer the audio quality of the samples. A test setup with concert recordings manually crawled from YouTube is used to validate the presented method. The results show that the proposed method achieves better results than previous methods.