- Thomas Hermann, Helge Ritter (2002)
In Nakatsu, R. and Kawahara, H. (Ed.) Proceedings of the International Conference on Auditory Display, p. 76--81, International Community for Auditory Display (ICAD), ICAD, Kyoto, Japan
[BibTeX Entry]
[Download PDF]
Summary
This paper introduces Crystallization Sonification, a sonification model for exploratory analysis of high-dimensional datasets. The model is designed to provide information about the intrinsic data dimensionality (which is a local feature) and the global data dimensionality, as well as the transitions between a local and global view on a dataset. Furthermore the sound allows to display the clustering in high-dimensional datasets. The model defines a crystal growth process in the high-dimensional data-space which starts at a user selected ``condensation nucleus'' and incrementally includes neighboring data according to some growth criterion. The sound summarizes the temporal evolution of this crystal growth process. For introducing the model, a simple growth law is used. Other growth laws which are used in the context of hierarchical clustering are also suited and their application in crystallization sonification offers new ways to inspect the results of data clustering as an alternative to dendrogram plots. In this paper, the sonification model is described and example sonifications are presented for some synthetic high-dimensional datasets.
- Table 1: Sound Examples for Crystallization Sonification for 5d Gaussian dataset
- Table 3: Sound examples for DCS on variation of the energy decay time
- Table 4: Sound examples for DCS on variation of the sonification time
- Table 5: Sound examples for DCS for different excitation locations
- Table 6: Sound examples for DCS for the mixture of a 2d uniform distribution and a 5d Gaussian
Contact
Thomas Hermann
Helge Ritter
Crystallization Sonification of High-dimensional Datasets
Sound Demonstrations for the Data Crystallization Sonification (DCS) Model
File/Track: | DCS started (E1a) at center, (E1b) in tail, (E1c) from far outside |
Description: | DCS for dataset that is N{0, I_5} excited at different locations |
Duration: | 1.4 s |
File/Track: | tau_(1/2) = (E3a) 0.001, (E3b) 0.005, (E3c) 0.01, (E3d) 0.05, (E3e) 0.1, (E3f) 0.2 |
Description: | DCS for a mixture of two Gaussians varying the energy decay time tau_(1/2) |
Duration: | 1.4 s |
File/Track: | T = 0.2s (E4a) , 0.5s (E4b), 1s (E4c) , 2s (E4d) , 4s (E4e) , 8s (E4f) |
Description: | DCS for a mixture of two Gaussians on varying the duration T |
Duration: | 0.2s -- 8s |
File/Track: | starting point: C0 (E5a), C1(E5b), C2(E5c) |
Description: | DCS for a mixture of three Gaussians in 10d space with different rang(Sigma) = {2,4,8} |
Duration: | 1.9 s |
File/Track: | condensation nucleus in (x0,x1)-plane at: E6a (-6,0)=C1, E6b(-3,0)=C2, E6c( 0,0)=C0 |
Description: | DCS for a mixture of a uniform 2d and a 5d Gaussian |
Duration: | 2.16 s |