Deep neural networks for small footprint text-dependent speaker verification

TLDR

The study investigates applying deep neural networks to small‑footprint text‑dependent speaker verification. A DNN is trained to classify speakers at the frame level, and its last hidden‑layer activations are averaged to form d‑vectors that serve as speaker models for enrollment and verification. The DNN‑based system outperforms the i‑vector baseline, achieving 14% and 25% relative EER improvements in clean and noisy conditions and showing greater robustness to additive noise, especially at low false‑rejection rates.

Abstract

In this paper we investigate the use of deep neural networks (DNNs) for a small footprint text-dependent speaker verification task. At development stage, a DNN is trained to classify speakers at the framelevel. During speaker enrollment, the trained DNN is used to extract speaker specific features from the last hidden layer. The average of these speaker features, or d-vector, is taken as the speaker model. At evaluation stage, a d-vector is extracted for each utterance and compared to the enrolled speaker model to make a verification decision. Experimental results show the DNN based speaker verification system achieves good performance compared to a popular i-vector system on a small footprint text-dependent speaker verification task. In addition, the DNN based system is more robust to additive noise and outperforms the i-vector system at low False Rejection operating points. Finally the combined system outperforms the i-vector system by 14% and 25% relative in equal error rate (EER) for clean and noisy conditions respectively.