ESDD 2026: Environmental Sound Deepfake Detection Challenge Evaluation Plan

2025-08-06

AI Summary

To address the lack of large-scale, diverse benchmark datasets for environmental sound deepfake detection (ESDD), this work introduces EnvSDD—the first high-quality annotated dataset comprising 45.25 hours of authentic and 316.7 hours of synthetic environmental audio, spanning multiple acoustic scenes and generative models. Building upon EnvSDD, we organize the ESDD 2026 International Challenge, featuring two realistic evaluation tasks: “unknown generator identification” and “low-resource black-box detection.” We further propose an acoustic-feature-driven deep learning framework and conduct systematic evaluations of model generalizability, robustness, and few-shot adaptability. EnvSDD fills a critical gap in ESDD benchmarking and significantly advances the practical deployment of audio forensic technologies.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Fact-Checking / Misinformation Detection (NLP Focus)
• Computer Vision: Diffusion Models for Vision

Application Category

• Web Mining and Content Analysis: Web data provenance, reliability, and authenticity
• Security and Privacy: Large-scale security measurements
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics

Abstract

Recent advances in audio generation systems have enabled the creation of highly realistic and immersive soundscapes, which are increasingly used in film and virtual reality. However, these audio generators also raise concerns about potential misuse, such as generating deceptive audio content for fake videos and spreading misleading information. Existing datasets for environmental sound deepfake detection (ESDD) are limited in scale and audio types. To address this gap, we have proposed EnvSDD, the first large-scale curated dataset designed for ESDD, consisting of 45.25 hours of real and 316.7 hours of fake sound. Based on EnvSDD, we are launching the Environmental Sound Deepfake Detection Challenge. Specifically, we present two different tracks: ESDD in Unseen Generators and Black-Box Low-Resource ESDD, covering various challenges encountered in real-life scenarios. The challenge will be held in conjunction with the 2026 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP 2026).

Problem

• Detecting deceptive environmental audio deepfakes for misuse prevention
• Addressing limited scale and diversity in current ESDD datasets
• Evaluating detection models in real-world black-box low-resource scenarios

Innovation

• Large-scale curated dataset EnvSDD
• Two challenge tracks for detection
• Integration with ICASSP 2026