End-to-End Losses Based on Speaker Basis Vectors and All-Speaker Hard Negative Mining for Speaker Verification

Citations Over TimeTop 10% of 2019 papers

Abstract

In recent years, speaker verification has primarily performed using deep neural networks that are trained to output embeddings from input features such as spectrograms or Mel-filterbank energies.Studies that design various loss functions, including metric learning have been widely explored.In this study, we propose two end-to-end loss functions for speaker verification using the concept of speaker bases, which are trainable parameters.One loss function is designed to further increase the interspeaker variation, and the other is designed to conduct the identical concept with hard negative mining.Each speaker basis is designed to represent the corresponding speaker in the process of training deep neural networks.In contrast to the conventional loss functions that can consider only a limited number of speakers included in a mini-batch, the proposed loss functions can consider all the speakers in the training set regardless of the mini-batch composition.In particular, the proposed loss functions enable hard negative mining and calculations of betweenspeaker variations with consideration of all speakers.Through experiments on VoxCeleb1 and VoxCeleb2 datasets, we confirmed that the proposed loss functions could supplement conventional softmax and center loss functions.

Related Papers

• → Speaker verification by inexperienced and experienced listeners vs. speaker verification system(2011)10 cited
• → Text-independent speaker verification using speaker clustering and support vector machines(2003)6 cited
• Text-Dependent Speaker Verification Implemented in Matlab Using MFCC and DTW(2010)
• → Using nonstandard SVM for combination of speaker verification and verbal information verification in speaker authentication system(2002)
• → Fixed-Phrase Speaker Verification(2011)