Supervised attention for speaker recognition
This work addresses a specific bottleneck in speaker recognition systems for applications like voice authentication, though it is incremental as it builds on existing attention-based methods.
The paper tackled the problem of self-attentive pooling underperforming compared to temporal average pooling in speaker recognition by introducing supervised training strategies for the attention mechanism, resulting in improved performance that outperforms existing methods in various settings, including short utterance recognition, and achieves competitive results on VoxCeleb datasets.
The recently proposed self-attentive pooling (SAP) has shown good performance in several speaker recognition systems. In SAP systems, the context vector is trained end-to-end together with the feature extractor, where the role of context vector is to select the most discriminative frames for speaker recognition. However, the SAP underperforms compared to the temporal average pooling (TAP) baseline in some settings, which implies that the attention is not learnt effectively in end-to-end training. To tackle this problem, we introduce strategies for training the attention mechanism in a supervised manner, which learns the context vector using classified samples. With our proposed methods, context vector can be boosted to select the most informative frames. We show that our method outperforms existing methods in various experimental settings including short utterance speaker recognition, and achieves competitive performance over the existing baselines on the VoxCeleb datasets.