DFKI-Speech System for WildSpoof Challenge: A robust framework for SASV In-the-Wild

TL;DR

The paper tackles SASV under real-world, noisy conditions by leveraging the SpoofCeleb dataset and the WildSpoof challenge to emphasize in-the-wild variability. It introduces a tandem architecture in which a spoofing detector built on a wav2vec 2.0 XLS-R front end with a top-3 Mixture-of-Experts fusion and a graph-based AASIST backend operates in parallel with a low-complexity ReDimNet-based ASV system that fuses multiscale 1D/2D features and is trained with SphereFace ($s=30$, $m=1.5$) and Circle loss (weight $0.2$), complemented by AS-Norm and model ensembling. Training uses SpoofCeleb with 1,250+ speakers and 23 spoofing systems, with extensive augmentation (MUSAN, RIR, RawBoost) and optimization on H100 GPUs. Results show substantial gains over baselines, achieving an a-DCF of $0.032$-level with ensembling on SpoofCeleb and a macro a-DCF of $0.2022$ on WildSpoof, demonstrating robust SASV performance in unconstrained environments and guiding practical deployments.

Abstract

This paper presents the DFKI-Speech system developed for the WildSpoof Challenge under the Spoofing aware Automatic Speaker Verification (SASV) track. We propose a robust SASV framework in which a spoofing detector and a speaker verification (SV) network operate in tandem. The spoofing detector employs a self-supervised speech embedding extractor as the frontend, combined with a state-of-the-art graph neural network backend. In addition, a top-3 layer based mixture-of-experts (MoE) is used to fuse high-level and low-level features for effective spoofed utterance detection. For speaker verification, we adapt a low-complexity convolutional neural network that fuses 2D and 1D features at multiple scales, trained with the SphereFace loss. Additionally, contrastive circle loss is applied to adaptively weight positive and negative pairs within each training batch, enabling the network to better distinguish between hard and easy sample pairs. Finally, fixed imposter cohort based AS Norm score normalization and model ensembling are used to further enhance the discriminative capability of the speaker verification system.

Figures (1)

• Figure 1: Proposed SASV framework architecture.