test_multiple_unheard on Replica environment.

facebookresearch / sound-spaces

A first-of-its-kind acoustic simulation platform for audio-visual embodied AI research. It supports training and evaluating multiple tasks and applications.

https://soundspaces.org

Creative Commons Attribution 4.0 International

322 stars 55 forks source link

test_multiple_unheard on Replica environment. #98

Closed chenjinyubuaa closed 1 year ago

chenjinyubuaa commented 1 year ago

Hello! I try to test the av_wan model on the Replica environment. But the test_multiple_unheard split does not exist on the provided dataset link. I tried the unheard model on val_multiple_unheard split, and the SPL is 26 which is far from the accuracy in the paper(34). So is the test_multiple_unheard split on Replica missing?

ChanganVR commented 1 year ago

Hi @chenjinyubuaa I'm not sure why the split was missing but I reuploaded it here: https://drive.google.com/file/d/1vpeCJp6ZJTXbCrgjXt8M74RH847ZlYBH/view?usp=sharing.

chenjinyubuaa commented 1 year ago

Thank you!