How to Improve Your Speaker Embeddings Extractor in Generic Toolkits

11/05/2018
by   Hossein Zeinali, et al.
0

Recently, speaker embeddings extracted with deep neural networks became the state-of-the-art method for speaker verification. In this paper we aim to facilitate its implementation on a more generic toolkit than Kaldi, which we anticipate to enable further improvements on the method. We examine several tricks in training, such as the effects of normalizing input features and pooled statistics, different methods for preventing overfitting as well as alternative non-linearities that can be used instead of Rectifier Linear Units. In addition, we investigate the difference in performance between TDNN and CNN, and between two types of attention mechanism. Experimental results on Speaker in the Wild, SRE 2016 and SRE 2018 datasets demonstrate the effectiveness of the proposed implementation.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
04/03/2020

Neural i-vectors

Deep speaker embeddings have been demonstrated to outperform their gener...
research
04/07/2021

Adapting Speaker Embeddings for Speaker Diarisation

The goal of this paper is to adapt speaker embeddings for solving the pr...
research
04/07/2020

Multi-Scale Aggregation Using Feature Pyramid Module for Text-Independent Speaker Verification

Currently, the most widely used approach for speaker verification is the...
research
02/28/2022

Magnitude-aware Probabilistic Speaker Embeddings

Recently, hyperspherical embeddings have established themselves as a dom...
research
04/26/2018

On deep speaker embeddings for text-independent speaker recognition

We investigate deep neural network performance in the textindependent sp...
research
04/06/2021

Speaker embeddings by modeling channel-wise correlations

Speaker embeddings extracted with deep 2D convolutional neural networks ...
research
05/19/2018

Sparse Architectures for Text-Independent Speaker Verification Using Deep Neural Networks

Network pruning is of great importance due to the elimination of the uni...

Please sign up or login with your details

Forgot password? Click here to reset