GitHub - YChenL/DS-TDNN: Official implement of "Dual-stream Time-Delay Neural Network with Dynamic Global Filter for Speaker Verification" in PyTorch · GitHub

Repository navigation

DS-TDNN

Public

About

Official implement of "Dual-stream Time-Delay Neural Network with Dynamic Global Filter for Speaker Verification" in PyTorch

arxiv.org/pdf/2303.11020v2.pdf

automatic-speaker-verification deep-learning nerual-network time-delay-neural-network

41 stars

3 watching

Report repository

Name		Name	Last commit message	Last commit date
Figs		Figs
Models		Models
exp/DS_TDNN_B/model		exp/DS_TDNN_B/model
loss		loss
modules		modules
visualization		visualization
Calculate.ipynb		Calculate.ipynb
MyModel.py		MyModel.py
README.md		README.md
dataLoader.py		dataLoader.py
requirements.txt		requirements.txt
train.py		train.py

Repository files navigation

Detials will coming soon. (Updating)

This repo includes implements of several SOTA front-end models for ASV

ECAPA-TDNN (Un-official)
SE-ResNet (Un-official)
MFA-TDNN (Un-official)
MFA-Conformer (Official)
DS-TDNN (Ours)

and some popular loss functions for ASV

Overviwes of DS-TDNN

Architecture

Experiment Results

About

Official implement of "Dual-stream Time-Delay Neural Network with Dynamic Global Filter for Speaker Verification" in PyTorch

arxiv.org/pdf/2303.11020v2.pdf

automatic-speaker-verification deep-learning nerual-network time-delay-neural-network

41 stars

3 watching

Report repository

Releases

Packages

Used by

Contributors

Languages