DATA EXPLORATION

Listening to our data

Datasets can be very large and complex. This means that standard visualisation approaches are often insufficient to represent the full complexities of the data. Sonification offers an alternative to explore the data. Check out some of our work using sonification to explore scientific datasets!

If you want to try this yourself you have a couple of options:

(1) Work directly with our Python package called STRAUSS, for which we've prepared example notebooks and code. This gives you access to the full potential of our sonification code. 

(2) Use the Sonification Suite, which is an online App and lets you produce some time-series type sonifications by uploading your own data or accessing some datasets directly through the App. This gives you a limited number of possibilities, but is designed to be an easier method to get started and still provide some customisation. 

Introducing COLIBRE's Sonified Interactive Maps

Team member James Trayford is also a developer COLIBRE simulations, a suite of simulations that track and and model how galaxies were formed. These are huge simulations, simulating 1000s of galaxies at a time and keeping track of a large number of properties of gas, stars, and dark matter, across space and time.  Check out the dedicated page on Sonification of the COLIBRE simulations, which shows some movies of how the galaxies evolve (with sonification included of course!) Along with the movies you will find on that page, we've developed an interactive tool to explore the data from these simulations. 

The sonified maps are avalaible to explore here.

These interactive maps combine sound and visuals. This means that multiple properties can be explored simultaneously at the same locations in the maps. By enabling sonification, users can start to uncover physics they don't see, simply by moving the cursor. For example, the unseen network of matter connecting galaxies in a 'cosmic web'. 

Can you hear the intensifying whoosh representing higher gas temperatures around galaxies? Or the fast ticking represent the high rate of X-ray radiation? Perhaps the evolving sound of a chord helps you find a clump of dark matter without any stars?

This tool is intended to give users freedom to explore: you can choose to combine and represent layers, however you like with our three sonification 'voices'. By building their own combination, users can help us to find new and better ways of applying this technique.

This interactive tool for these galaxy simulations is just the start! The plan is to develop these approaches and extend them to other areas of science and walks of life through the Audio Universe. 

The video below shows the explorer in action, with the user choosing various different layers to represent as sound. 

We have been developing approaches to turn spectra into sound. Light spectra, can be directly, and intuitively be turned into a sound spectrum. The result is that the features in the light spectrum at different wavelengths correspond to the different frequencies that make up the corresponding sound. We introduced this approach in Trayford et al. 2023. This example is a galaxy spectra containing a supermassive black hole. The bright features in the spectra (emission lines) correspond to different transitions associated with different chemical species. Each features results in a specific tone that can be heard in the sound. You can try this out for yourself below in the browser-based notebooks.

Our team member, Rose Shepherd, has been taking this approach to a much larger scale - exploring how we can use this method to identify and characters growing black holes in giant spectroscopic datasets. The analysis has already been performed, so watch this space for the associated research publication!

This mock data, shows how we can turn time-series day into sound. In this example the "brightness" values in the data are mapped to a cut-off frequency applied to a musical chord. The result of this is the sound appears to be "brighter" for higher (brighter) data values and duller for lower data values. More detail is presented in Tucker-Brown et al. 2022. 

This is just one of the many approaches to sonification that can be achieved with STRAUSS. 

Team member Rose Shepherd has done extensive research in using sonification to represent light curve data for public communications. The paper is under review, so watch this space to see the results!

You can also sonify timeseries data easily using our online App: Sonification Suite.

The code to make a variety of different approaches to sonifying timeseries data is also available to you on our STRAUSS page.

With our code you can listen to multiple variables at the same time! This can be particularly useful if you want to understand how different variables relate to one another. In this example, from Trayford et al. 2023, we can listed to (1) the star formation rate (SFR) and (2) the ‘metal’ mass fraction (known as ’stellar metallicity’), of a simulated galaxy from the EAGLE simulations over 13 billion years of cosmic time. The metallicity is mapped to a low-pass filter cut off frequency (see right axes), and we provide two different example mappings for simultaneously mapping the star formation rate to volume low frequency oscillator frequency, as represented in the axes.

The code to make this example is available to you on our STRAUSS page.