DeepAI AI Chat
Log In Sign Up

End-to-End Residual CNN with L-GM Loss Speaker Verification System

by   Xuan Shi, et al.
Shanghai University

We propose an end-to-end speaker verification system based on the neural network and trained by a loss function with less computational complexity. The end-to-end speaker verification system consists of a ResNet architecture to extract features from utterance, then mean pool to produces utterance- level speaker embeddings, and train using the large-margin Gaussian Mixture loss function. Influenced by the large-margin and likelihood regularization, large-margin Gaussian Mixture loss function benefits the speaker verification performance. Experimental results demonstrate that the Residual CNN with large- margin Gaussian Mixture loss outperforms DNN-based i-vector baseline by nearly 10


page 1

page 2

page 3

page 4


Large Margin Softmax Loss for Speaker Verification

In neural network based speaker verification, speaker embedding is expec...

Generalized End-to-End Loss for Speaker Verification

In this paper, we propose a new loss function called generalized end-to-...

Dr-Vectors: Decision Residual Networks and an Improved Loss for Speaker Recognition

Many neural network speaker recognition systems model each speaker using...

Generative Adversarial Speaker Embedding Networks for Domain Robust End-to-End Speaker Verification

This article presents a novel approach for learning domain-invariant spe...

Rethinking Feature Distribution for Loss Functions in Image Classification

We propose a large-margin Gaussian Mixture (L-GM) loss for deep neural n...

Supervector Compression Strategies to Speed up I-Vector System Development

The front-end factor analysis (FEFA), an extension of principal componen...