Differentiable Supervector Extraction For Encoding Speaker And Phrase Information In Text Dependent Speaker Verification

Abstract

In this paper, we propose a new differentiable neural network alignment mechanism for text-dependent speaker verification which uses alignment models to produce a supervector representation of an utterance. Unlike previous works with similar approaches, we do not extract the embedding of an utterance from the mean reduction of the temporal dimension. Our system replaces the mean by a phrase alignment model to keep the temporal structure of each phrase which is relevant in this application since the phonetic information is part of the identity in the verification task. Moreover, we can apply a convolutional neural network as front-end, and thanks to the alignment process being differentiable, we can train the whole network to produce a supervector for each utterance which will be discriminative with respect to the speaker and the phrase simultaneously. As we show, this choice has the advantage that the supervector encodes the phrase and speaker information providing good performance in

Stats

• citations1
• S2 citations—
• github stars0
• HF likes0
• heat score2.26
• arxiv keymingote2018differentiable

Related papers

• Optimization Of The Area Under The ROC Curve Using Neural Network Supervectors For Text-dependent Speaker Verification (2019)10.21
• End-to-end Attention Based Text-dependent Speaker Verification (2017)14.87
• An End-to-end Text-independent Speaker Verification Framework With A Keyword Adversarial Network (2019)5.84
• Speaker Verification Using Convolutional Neural Networks (2018)0.00
• Adapting End-to-end Neural Speaker Verification To New Languages And Recording Conditions With Adversarial Training (2018)9.59
• Exploring The Use Of An Unsupervised Autoregressive Model As A Shared Encoder For Text-dependent Speaker Verification (2020)5.84
• Deep Speaker Embedding Learning With Multi-level Pooling For Text-independent Speaker Verification (2019)0.00
• Y-vector: Multiscale Waveform Encoder For Speaker Embedding (2020)8.60