EnvSSLAM-FFN: Lightweight Layer-Fused System for ESDD 2026 Challenge

TL;DR

EnvSSLAM-FFN is a lightweight, layer-fused system that effectively detects environmental sound deepfakes under challenging conditions, outperforming official baselines in the ESDD 2026 Challenge.

Contribution

The paper introduces a novel fusion of SSLAM self-supervised encoder layers with a lightweight FFN for improved deepfake detection in environmental sounds.

Findings

Achieved low EERs of 1.20% and 1.05% on two challenge tracks.

Outperformed official baselines consistently.

Effective in severe data imbalance scenarios.

Abstract

Recent advances in generative audio models have enabled high-fidelity environmental sound synthesis, raising serious concerns for audio security. The ESDD 2026 Challenge therefore addresses environmental sound deepfake detection under unseen generators (Track 1) and black-box low-resource detection (Track 2) conditions. We propose EnvSSLAM-FFN, which integrates a frozen SSLAM self-supervised encoder with a lightweight FFN back-end. To effectively capture spoofing artifacts under severe data imbalance, we fuse intermediate SSLAM representations from layers 4-9 and adopt a class-weighted training objective. Experimental results show that the proposed system consistently outperforms the official baselines on both tracks, achieving Test Equal Error Rates (EERs) of 1.20% and 1.05%, respectively.