Today’s prompt for the 12 Days of AI has to do with voice and audio manipulation, and intellectual property (who owns your voice? Who should have access to your voice? Are YOU your voice?) and as AI gets more powerful, this voice replication will become more realistic, with lots of ramifications around identity.
The prompt, however, had us playing with Holly+, an audio experiment from the artist Holly Herndon (I tinkered with the site a year ago — see my post about my inquiry). We are asked to record our voice, and use the site, allowing Holly+ to manipulate the audio file in its own idiosyncratic way that folds the audio and wraps it in odd effects.
I recorded myself saying versions of: Holly, can you hear me? and then listened to the audio transformation. I took the audio file from Holly+ and brought it into a music mixer and layered some synth sounds under it, to make it into a sort of piece of music. I also added two versions of my original file and then played around with some echo effects.
One of the inquiry questions revolves around intellectual property and whether one’s voice can be used with consent, and I say, no, but given how many of us have recorded audio, video and other media, it seems nearly impossible to really know if any AI platform has used our voice to train its system. Holly+ is built on privacy, of not retaining any of the uploaded voice recordings, but who really knows, right?
It’s a strange world, but I appreciate artists like Herndon pushing the boundaries of digital art, and creating a space for experimentation.
But, Holly, did you hear me?
Peace (in the sound void),
Kevin