To record bat echolocation and visually analyze the waveforms to determine the species and type of call. Different species echolocate at different frequencies and the calls they make when they're travelling, searching for prey and diving to eat all look different from each other and different between species.
With the improvements to machine translation without parallelism data and voices-to-voice translation, how stupid is the idea that we might someday be able to take enough bat calls and human speech, and create a true translation system? obviously assumes the bats got something vaguely resembling a language we can translate from.
Who would think that an animal would just make random noises?
That just ridiculous.
"Nah! It was your turn you big bully!"