The complete speaker clustering task can be separated into two components. The first one is audio segmentation based on speaker changing points. The second is putting the identified voice utterances into groups. Here a particular group should contain utterances from a single speaker. These components include the phases of data preprocessing, feature extraction, segmentation, and clustering. For this project, the current best method will be used for the segmentation part and the research focus is provided to the clustering component especially in the feature engineering domain.
Team Members - S.Ketharan
