מרק לסט

Timestamped observations, generally known as time series, may contain valuable information about a variety of natural and man-made phenomena ranging from weather changes to stock markets. Our capability to collect such data has increased dramatically due to advances in computing and sensory technologies. Visualization is known as a very effective tool for interactive data exploration tasks. In this research, we have tested the hypothesis that musical sonification (the use of musical audio) can serve as a viable alternative to visualization of time-series data whenever the visual representation is unavailable or impossible to use. We have developed a time-series sonification technique, which utilizes some important features of Western tonal music to convert univariate and multivariate time series into a musical equivalent. The technique was used to conduct two online user studies, where the subjects were asked questions about the data behavior by listening to a musical display of time series rather than viewing their visual representation. The results of both studies indicate that our methodology for musical representation of time-dependent data allows most users, including people with low musical hearing ability, to successfully perform a variety of common data exploration tasks.