Is there some relation between audio input and motion input? Could you please provide some inference code for other custom audio inputs?

[liruilong940607]

Hi We are currently not support that in the code base. But it should be that hard to modify the code.

Please to refer to the preprocessing code on how to extract audio features and create the test file:

https://github.com/google-research/mint/blob/main/tools/preprocessing.py