A multi-level web-based visualization of sound collections

Sound collections are becoming more frequently used in our daily lives. A common dilemma is how to present a large sound collection in a concise way but at the same time enable user access to all relevant data. Main objective of this thesis is creation of the system for concise visualization of collections of sound data. To achieve that we have used a two-level methodology for visualization, which means we can move between data clusters on two levels. Sound records are clustered based on similarity between them. For computation of similarities we have used attributes that are retrieved from Essentia library. The advantage of using our methodology is the ability to view data in a manageable way and research the data collection in greater detail due to the multi-level design