[notice] Before leaving an issue

[SaoYear]

Hi everyone,

Thank you for your interest in ATST-SED.

Since there is only me to maintain this repo, it is hard to provide intermediate response to everyone. Before posting the issue, please check the following things:

1. The coding structure of ATST-SED is based on the DESED repo. So, if you are a new-comer to DESED/DCASE challenge, please make sure you understand the DESED codes.

2. The version changes of torchmetrics and pytorch-lightning packages could lead to problem in reproduction of ATST-SED. Here is an environment list for your reference (python=3.9.X):

scipy==1.7.3
sed_scores_eval==0.0.0
SoundFile==0.10.3.post1
thop==0.1.1.post2209072238
torch==1.12.0+cu116
torchaudio==0.12.0+cu116
torchmetrics==0.11.4
tqdm==4.62.3
pytorch-lightning==1.5.1

4. After checking the above two points, if you are facing any error - please provide the full terminal loggings; if you having problem on understanding any step of ATST-SED, feel free to leave an issue.

5. I will response the issues in every Monday.

Again, thanks for everyone's attention on our work and please understand me if I did not response you in time.

• If you cannot fix your problem anyway, for Chinese users, you could contact me on wechat: ShaoNian_ishere. For other oversea users, you could contact me by: shaonian@westlake.edu.cn .