===== Audio =====
  * [[https://dagshub.com/kinkusuma/esc50-dataset|Environmental Sound Classification]]: 2000 environmental audio recordings suitable for benchmarking methods of environmental sound classification.
  * [[https://drive.google.com/drive/folders/0By0bAi7hOBAFUHVXd1JCN3MwTEU?resourcekey=0-ET4Q9_rMn796b9mdO5kcKA|voices?]] or might be environmental sounds
  * [[https://www.openslr.org/12|1000 hours of English speech]]
  * [[https://www.analyticsvidhya.com/datahack/contest/practice-problem-urban-sound-classification/]]