@yi really interested in the talk though. is there any chance you could extract the audio, normalise it with audacity, and re-upload the fixed audio?