On this website, you can find detailed results, examples, and the code corresponding to our paper.
In Uncover CleverHans, you will find examples similar to those presented in the paper, but with accompanying audio for listening.
In Benchmarking, you will find detailed results on hyperparameter optimization, a comparison between metrics (AUC and FF), and examples for each dataset. We recommend reading them in the following order: Drums → AudioSet → KWS, as each section builds upon the previous one. However, they can also be read in any order, though some comments may not make sense without prior context.
The code to replicate our results or test on other datasets can be found here.
The models used can be found here, except for AST, which is publicly available on Hugging Face.
Synthetic datasets can be found here.
We plan to continuously expand the code and this website by adding new datasets and systems as our research progresses.
Note: Our paper is currently under review in the anonymity period. Once the review process is complete, we will provide a link to the paper.