a video game designer and a heartland conservative by akstuhl published on 2019-03-09T03:38:47Z a test of a work-in-progress software instrument called streameddies, where snippets of an incoming signal are processed by an audio classifier and then regurgitated semi-randomly according to which descriptors (e.g. "musical instrument" or "narrator, monologue") have come up most often. here, the process listened to the "lofi hip hop radio - beats to study/relax to" livestream and part of a TED talk. I steered the streameddies output by filtering out the speech-related descriptors at the start of playback, allowing these in, then filtering out all descriptors besides "heart murmur" and "heart sounds, heartbeat" toward the end.
