Testing Emodb model on audio files.

Hello, I want to test the pretrained models on the audio files, however i was able to test it from the EMODB.npy file that is provided there, but I was not able to convert any audio file to respective .npy file. can you provide me a method to test a single audio file through your pretrained model?