search

nesl / asvspoof2019

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

block is too much？ #7

Open terrymaoterui opened 4 years ago

terrymaoterui commented 4 years ago

The dimension of MFCC feature extraction is 72126 -- according to the official website data set and your feature extraction code It goes through six blocks The dimensions become 1 1 Because each block will be reduced by 3 times in length -- according to your paper, Deep Residual Neural Networks for Audio Spoofing Detection This does not match the number of nodes in the fc layer How to solve it？ thank you

wangziqi000 commented 4 years ago

Hi, Please refer to https://github.com/nesl/asvspoof2019/issues/9#issue-592336636