Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

DPRNN in WHAM! #55

Open
261145 opened this issue Sep 6, 2023 · 2 comments
Open

DPRNN in WHAM! #55

261145 opened this issue Sep 6, 2023 · 2 comments

Comments

@261145
Copy link

261145 commented Sep 6, 2023

你好,我想问一下您有跑过DPRNN在WHAM!数据集上的结果吗?
我看到别人论文里面报道的DPRNN在WHAM!数据集上的性能能接近14db。但是我自己跑的只有11db不到,不清楚是不是数据集的问题。 (参数设置应该没有问题)我用您的代码在WSJ0-2MIX数据集上的结果是正常的性能。希望您能指点一下

@JusperLee
Copy link
Owner

有两个可能得原因:

  1. 很多人使用的并不是原始dprnn实现。因为像Asteroid实现的是FasNet版本的dprnn。
  2. 有可能是lr decay的原因,当使用连续10个epoch loss不下降减少一半的lr,会提升分离性能。

@261145
Copy link
Author

261145 commented Sep 7, 2023

感谢,我先试试

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants