Graph attentive feature aggregation for text-independent speaker verification

TL;DR

This paper introduces a graph attentive feature aggregation module that models pairwise relationships between frame-level features for improved text-independent speaker verification, demonstrating consistent performance gains.

Contribution

It proposes a novel graph attention-based module for aggregating frame-level features, directly modeling inter-feature relationships for speaker verification.

Findings

Achieves over 10% relative improvement on baseline systems.

Effectively models pairwise feature relationships using graph attention.

Demonstrates versatility across different architectures and input features.

Abstract

The objective of this paper is to combine multiple frame-level features into a single utterance-level representation considering pairwise relationship. For this purpose, we propose a novel graph attentive feature aggregation module by interpreting each frame-level feature as a node of a graph. The inter-relationship between all possible pairs of features, typically exploited indirectly, can be directly modeled using a graph. The module comprises a graph attention layer and a graph pooling layer followed by a readout operation. The graph attention layer first models the non-Euclidean data manifold between different nodes. Then, the graph pooling layer discards less informative nodes considering the significance of the nodes. Finally, the readout operation combines the remaining nodes into a single representation. We employ two recent systems, SE-ResNet and RawNet2, with different input features and architectures and demonstrate that the proposed feature aggregation module consistently shows a relative improvement over 10%, compared to the baseline.