Devlog #42
Hello friends! Welcome to the Loud Numbers devlog - a weekly email about how we (that's Duncan Geere and Miriam Quick) are writing, composing, editing and producing the world's first podcast series turning data into music - a process called sonification. You're getting this email because you signed up at loudnumbers.net.
What's been happening this week?
Editing and Mastering
The main job for this week has been to get all of our scripts together and polished in preparation for our main recording session next week. Four of our five tracks are mastered and ready to go, courtesy of the geniuses at Queer Ear Mastering. So now we just need to record our voices explaining the stories and sonification systems.
Explaining the sonification systems means grabbing little samples of all of the tracks so you can hear what we're talking about when we're walking you through how each track works. It's like in Song Exploder, when you hear all the elements of the track before you hear the final thing all together. That's what Miriam has been focusing on.
Duncan, on the other hand, has been working through all of our scripts and just making sure everything is solid and still makes sense, especially for the scripts that we wrote a while ago. Assuming all goes well, then by next week we'll be set to record everything.
Then we've just got to edit it all - the voice tracks, the audio samples, the backing tracks, our theme tune, and the final track - together. That'll be the main job before release.
Handling Spiky Data
We've also been doing a little bit of work on another project for a client (you can hire us to do sonification work for you!). We can't talk about it much yet, but we did hit on a handy solution to the problem of mapping data to volume when the data you're using has a lot of outliers.
Normally this would be very difficult - you hear nothing most of the time, and then suddenly there's something very loud. But we hit upon another approach which combines two mappings in one.
Here's how it works. First you figure out the "normal" range of the data, ignoring the outliers. You map that to volume, and clip it so that values that exceed the normal range are just mapped to 100% volume. Then you calculate the amount that range is exceeded by for the outliers, and map that to something that modifies the character of the sound - distortion, or a filter, or something else. With a bit of slew to smooth it out, you get a really nice effect where you can hear the normal range, but also hear when something is dramatically exceeding the normal range because the timbre of the sound changes.
Try it out and see how you get on.
Elsewhere on the web
Another quiet week for sonification, but we did spy the following:
-
Yuri Suzuki's Welcome Chorus is an interactive installation of 12 interactive horns, inviting visitors into the Ars Electronica gallery space.
-
The Data Sonification Archive has done an analysis of its metadata, showing both subject matter and goal. Art is the most popular goal, and Earth Sciences are the most popular subject matter.
-
Radio New Zealand has a great-looking new series on the history of electronic music.
-
In 1973 Bhutan issued a set of postage stamps which were tiny, playable vinyl records. Adorable!
What have you seen or created lately? Mail us!
That's all for this week.
xox
Duncan & Miriam