This chapter introduces a concept of granular time series—models of time series in which information granules play a central role. Information granularity is instrumental in arriving at human-centric models of high level of interpretability and transparency. The chapter considers description, interpretation, and develops classification of time series by showing how a layered architecture of granular time series is developed. It elaborates the fundamental hierarchically organized layers of processing supporting the development of granular time series, namely, granular descriptors used in the visualization of time series, construction of linguistic descriptors used afterward in the generation of a linguistic description of time series. The notion of information granularity plays a pivotal role in all interpretation and analysis pursuits of time series. The chapter deals with an overall view of the conceptual framework stressing its key functionalities and a layered architecture and then moves on to a detailed discussion by elaborating on the supported algorithmic aspects.