We know. Unfortunately, the only thing we (the contributors) can do is to disable the audio for audio exercises. (and forward it to the higher powers who may decide to do something about it but I wouldn't really count on it in the near future. ;) Sadly, disabling the audio exercises doesn't mean disabling the audio altogether. We apologize for the inconvenience.
There are courses with human speakers (I am aware of the Esperanto course which has them). I realize that human voice recordings consume more memory / bandwidth but wouldn't it be worthwhile?