Audio Description Prototypes for Data Nuggets graphics

I’ve been a fan of Audio Descriptions (AD) in film and TV for awhile now. Most Netflix original content includes audio descriptions and I used to listen to more shows than I watch. I never have seen Daredevil, but I listened to every minute of it. And Stranger Things never sounded so good — and I still don’t know what the monster looks like. For those that want to listen and learn about the technology, there’s a good podcast on audio descriptions. Host Dallas Taylor describes how audio descriptions “could be for the masses,” and I agree. Audio descriptions are not simply for the blind and visually impaired. Just look at how podcasts have become a regular part of our media diet. Why not listen to our data visualizations and graphs too?

While audio descriptions are not sonifications, I wanted to explore the idea of creating audio descriptions as a baseline for sounding a graph when sonification isn't possible. I’m not a trained audio describer, so I may have conflated the “say what you see” rule with some science behind the graph a few times. As a document of the process, I created a script for all audio descriptions, recorded all audio descriptions in my home studio with a Sennheiser MKH-416 microphone, and edited and mixed the audio inside Logic Pro X with a little bit of EQ, compression, and de-essing (FabFilter Pro-Q3; Pro-C2; Pro-DS; and a bx_console SSL 9000J channel strip).

Below includes graphs from all five Data Nuggets our team has selected as part of the NSF AISL pilot. I’ve included only the graphics of data we are using for our sonification work. I copied most if not all of the audio description scripts as the alt-text for the graph images. I placed the matching audio description as .mp3 files next to each graph.