search

YChenL / DS-TDNN

Official implement of "Dual-stream Time-Delay Neural Network with Dynamic Global Filter for Speaker Verification" in PyTorch
https://arxiv.org/pdf/2303.11020v2.pdf
39 stars 7 forks source link

Parameter issue #1

Open 12-zhx opened 1 year ago

12-zhx commented 1 year ago

Why do you set the value of "T" in the global branch as 200? But length of speech is variable, the value is still 200 in the test phrase.

YChenL commented 10 months ago

Although the filter length is fixed, we can easily obtain filters with different resolutions by interpolating to handle the variable-length inputs during inference. A more detailed explanation is provided in Sec. II.B of the arXiv version of the paper.