nesl / asvspoof2019

Our submission to the ASVspoof 2019: Automatic Speaker Verification Spoofing and Countermeasures Challenge
96 stars 40 forks source link

Recurring GAP #20

Open potato-asuteed-potato opened 2 years ago

potato-asuteed-potato commented 2 years ago

Dear Author,

After running experiments on PA with your code, we got similar results to yours for both spect and mfcc features, and then a large EER gap between your reported results and our implementation when using cqcc features. Is there anything you need to pay attention to and change when using cqcc features?

Thank you for your reference. I really hope to get your answers。

Thanks again!