Data Sonification

Although it would likely be a waste of time, I have always been tempted to somehow turn geologic data I have collected into sound. It is commonplace to convert data into visual representations by which qualitative patterns are evaluated by simply looking at it (e.g., maps). Less common is converting data into sound and listening for patterns. The idea of turning data into sound for analysis is certainly not new (e.g., the Geiger counter was invented in the first decade of the 20th century), but the application of sonification to varied data types across scientific disciplines has seen more activity in the last couple of decades.

It is important to make the distinction between music and sound. A separate, but related, discussion would deal with the potential for creating new kinds of avant-garde music within this realm (check out here [1] and here [2]), but that's not what I'm talking about. Data sonification is simply the auditory equivalent of data visualization. Yes, some visual data is certainly striking in its beauty, but aesthetic value and scientific value are mutually exclusive. In recent years, data sonification has been applied especially in biomedical disciplines (e.g., here and here), but what about the geosciences?

What data types in geoscience would be amenable to being listened to? One answer to why we haven't seen the value of data sonification demonstrated in this regard is because it has none. This approach can most easily be applied with time series data (and potentially 'thickness series' data as I discuss below) but what about spatial information? My intuition is that turning spatial data into something that one can listen to would create something extremely difficult to interpret.

The two figures below show a waveform (top) and associated sonogram (bottom) for an eruption event at Mt. Etna (click on the image to go to the webpage for those figures, here for the home page [3], or here for a BBC report about the project in general [4]).

I listened to some of the data from the above project and I have to say I was underwhelmed. It should be noted, however, that I am currently sitting in a room where clicking, whooshing, and otherwise white-noiseing of a magical machine that knows when a zircon was born is overpowering all other sound. Additionally, this particular dataset is a gradual and subtle crecsendo...all-in-all, not a very exciting pattern. That being said, one of the objectives of what they are doing is to figure out a workflow.


Being a sedimentary geologist, I measure and describe a lot of vertical sections of stratigraphic successions that are exposed in outcrop or part of wellbore data (e.g., core, logs) for the subsurface. At right is an example of a measured section from outcrop. How might we go about sonifying this? From my perspective, I would want attributes such as lithology, grain size, bed thickness, prominent sedimentary structures, presence/absence of trace fossils, and so on captured in the data. The question is: would doing this add any value to the process of recognizing and then interpreting observed (heard) patterns?

Perhaps the value might be when multiple sections are combined? I don't know. This is one of those topics that is perfect for a blog because I'd like to hear what people think. Do you think your particular type of data could be converted to sound and be of any use?





Further reading (and some listening) on the web:
(1) Article from Feb 2007 SF Chronicle about Stanford music professor turning data into sound
(2) Heart Chamber Orchestra
(3) Mt. Etna Volcano Sonification: Listening to Seismograms
(4) BBC News report about Mt. Etna project
(5) A Spatialized Meterological Data Sonification Project (PDF)