Re: [spam][crazy][fiction][random] Non-Canon MCBoss Spinoffs

Here’s from paperswithcode.com (which tends to be a little out of date) a state-of-the-art machine learning paper on audio source separation from august 9th: https://paperswithcode.com/paper/separate-anything-you-describe Roughly, they have a large model they trained that can separate typed prompts from recordings, as well as perform other tasks. This seems like it could be helpful. To identify where a sound is coming from, one usually has multiple channels. Wonder if it does that, or what does. It’s hard for me to consider and integrate multiple approaches nowadays. I feel like I could if I could get over some severe cognitive tension.