video in AudioSet
A list of about a million ten-second YouTube clips, each labelled with the sounds in it, that Google published in 2017. Labs use it to train and test general audio and music models.
Including "Mean Girls, Meaner Women" author featured on KTLA-TV in Los Angeles