Initial Ideas

regional selection

not all data needs to sound

look for correlations

ranges: where should we look for correlation

how to select

spatial dimensions by stream

select peaks of rainfall, play only those who

what advantages does sonification bring here?

• pattern recognition
• sound image, no better: sound object
• we might be in grip of the image

problem: all data is too much

rhythms in 3D

check for patterns first without elevation, maybe later compare.

maybe cursor pointer on one location -> time series on two channels, one for memory/comparison, one for interaction

maybe good to only compare one spot.

-> combination would be an abstract navigation scheme stream based.

pattern recognition often uses random data, so why not use a random walk?

we need to cache some data

restricted random walk

create one sonic cluster from one trace

Experimenting

testing how fast we can load single data points - ok., faster that "real" time

one slice through time

Evaluation

comparison of different regions would be valuable, mean should be an area

Whats next?

fuzzy indexing into data (averaging)

modulating averaging range

fuzzy in 3 dimensions