Devlog #32
Hello, hello, hello and welcome to everyone who’s joined us in the last couple of weeks! We are Duncan and Miriam, and this is the Loud Numbers development log - a weekly newsletter giving a behind-the-scenes glimpse into how we’re building the world’s first data sonification podcast.
Where It’s At
We’ve not done a big-picture catch-up in a while, so here’s the rough status on all the episodes that’ll make up season one of Loud Numbers.
- Boom & Bust - music complete, script complete. Done! This is the one we played at Outlier.
- The Natural Lottery - music was complete but we’re revisiting it, script complete.
- Tasting Notes - music complete, script pretty much done. Interviewing our guest star this weekend, then it’ll be done.
- The End of the Road - music complete, script not started.
- Europe (placeholder title) - music partially done, script not started.
- Brexit (placeholder title) - music partially done, script not started.
So we’re maybe three quarters of the way there? Especially as the scripts are much easier to write than the music. We’re aiming to close out the rest of the tracks in the next month or two, then get the scripts written and recorded after that. Once that’s all done we’ll finally be able to start thinking about a release date.
Fast numbers
We’ve recently been experimenting with using stepped tempo changes in sonification. That means that the music speeds up when the data values are larger and slows down when they’re smaller.
Here’s an example from one of The Natural Lottery - one of our work-in-progress tracks on climate change. The data is grouped into eight-year time blocks, and the tempo changes in a stepwise fashion, between 100 and 150 bpm. A faster tempo means winter ice melted earlier in that eight-year period.
But how successful is this as a way of communicating data? We’re not convinced it works. Sudden tempo changes are pretty disorientating. Also, they create an unevenly stretched time grid which feels unpleasant, a bit like unevenly spaced grid lines on a chart. So we’re going to try another approach for now!
Would love to hear some examples of data sonifications that do use tempo changes successfully – anyone know of any? Hit reply if you do and let us know.
Sonic variables
In 1967, French cartographer Jacques Bertin developed the concept of visual variables in his book, Semiologie Graphique. These are a menu of things you can use to map data onto in a visualization - position, size, shape, value, color, orientation, and texture.
In a similar way, there are sonic variables - ways in which sound can vary - which can be mapped to data. Pitch For example pitch: the higher the pitch, the larger the data value. In our recent Outlier talk (video coming soon), we presented the following list of sonic variables:
- Pitch (higher equals more)
- Loudness (louder means more)
- Number of instruments (more instruments playing equals more data)
- EQ (bassier equals more)
- Duration (longer sound equals more)
- Number of sounds in a time period (more sounds equals more)
- Tempo (faster equals more)
But this isn’t the only list. Researchers have been trying to classify sonic variables for a while. For example, JB Krygier in 1994, cited in this more recent paper by Florian Hruby, listed the following nine sound variables: location, loudness/volume, pitch, register, timbre, duration, rate of change, order and attack/decay.
Perhaps some of our variables could be grouped together. EQ and instruments into a single timbre variable, for example, or tempo subsumed into duration. But we think there’s also value in splitting them out into these more top-level and musically useful categories.
But there’s a whole world of sonic variables out there. And as we said in our Outlier talk, each one has unique qualities and distinct limitations. For example, loudness can have a powerful effect but is affected by the playback system (especially volume level and bass response) and also by audio engineering techniques like compression. The fun bit comes in working out which variables most effectively communicate different kinds of data.
Any good lists of sonic variables you know of? Again - hit reply.
More Than This
Elsewhere on the web, we’ve been peeping (what’s the sonic equivalent of peeping?) a pair of fantastic COVID sonifications. The first is made by the BBC and blends viz and sonification approaches. A beautiful piece.
Then the second is a remix of an NYT visualization by Carni Klirs. The double encoding of the number of cases growing over time is particularly effective. Really nice work.
That’s all we’ve got for you today. We’ll be interviewing our beer expert tomorrow - so we’ll let you know how it went next week. Catch you then!