Home

Awesome

Densely Connected Time Delay Neural Network

PyTorch implementation of Densely Connected Time Delay Neural Network (D-TDNN) in our paper "Densely Connected Time Delay Neural Network for Speaker Verification" (INTERSPEECH 2020).

News

Pretrained Models

We provide the pretrained models which can be used in many tasks such as:

D-TDNN & D-TDNN-SS

Usage

Data preparation

You can either use Kaldi toolkit:

Or checkout the kaldifeat branch if you do not want to install Kaldi.

Test

python evaluate.py --root $datadir/test_no_sil --model D-TDNN --checkpoint dtdnn.pth --device cuda

Evaluation

VoxCeleb1-O

ModelEmb.Params (M)LossBackendEER (%)DCF_0.01DCF_0.001
TDNN5124.2SoftmaxPLDA2.340.280.38
E-TDNN5126.1SoftmaxPLDA2.080.260.41
F-TDNN51212.4SoftmaxPLDA1.890.210.29
D-TDNN5122.8SoftmaxCosine1.810.200.28
D-TDNN-SS (0)5123.0SoftmaxCosine1.550.200.30
D-TDNN-SS5123.5SoftmaxCosine1.410.190.24
D-TDNN-SS1283.1AAM-SoftmaxCosine1.220.130.20

Citation

If you find D-TDNN helps your research, please cite

@inproceedings{DBLP:conf/interspeech/YuL20,
  author    = {Ya-Qi Yu and
               Wu-Jun Li},
  title     = {Densely Connected Time Delay Neural Network for Speaker Verification},
  booktitle = {Annual Conference of the International Speech Communication Association (INTERSPEECH)},
  pages     = {921--925},
  year      = {2020}
}

Revision of the Paper

References:

[16] X. Li, W. Wang, X. Hu, and J. Yang, "Selective Kernel Networks," in IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2019, pp. 510-519.