Grand Challenge 1: ESDD2 | ICME 2026 Session Schedule

1. Opening Remarks

9:30 – 9:40 AM · 10 min

Host: Ming Li

Welcome and introduction
Overview of the challenge/session
Introduction of schedule and presentation format

2. Invited Talk & Q&A

9:40 – 10:40 AM · 60 min

Speaker: Xin Wang Host: Ming Li

On Evaluation Metrics of Speech Deepfake Detection – Lessons Learned from ASVSpoof5

Abstract ASVspoof challenges, which are dedicated to speech anti-spoofing or deepfake detection, have been using discriminative-aware evaluation metrics in the past ten years. While they are useful in measuring how good the detectors are at discriminating real and fake data, they rely on an oracle setup and ignore the goodness of the detectors from the user's point of view. The latter aspect is also known as score calibration. This talk serves as a mini-tutorial on this missing aspect and shares the findings from the latest ASVspoof 5 challenge.

Bio Xin Wang is a Project Associate Professor and a JST PRESTO researcher at the National Institute of Informatics, Japan. He is one of the organizers of the last three ASVspoof Challenges. He is also a member of the organizing team for the last three VoicePrivacy Challenges, an international joint effort on speech privacy.

3. Oral Presentation Session

3.1 Challenge Summary Paper

10:40 – 11:00 AM · 20 min

Speaker: Xueping Zhang

Paper ID	Paper Title
223	Overview of ESDD2: Environment-Aware Speech and Sound Deepfake Detection Challenge

3.2 Coffee Break

11:00 – 11:10 AM · 10 min

3.3 Accepted Paper Spotlight Presentations

11:10 AM – 12:00 PM · 50 min

Format: 7 min each (5 min presentation + 2 min Q&A)

Certificate presentation and commemorative photos for each team; sponsor acknowledgement during the award presentation for the first-place team.

Order	Paper ID	Paper Title
1	208	Component-Level Ensemble Fusion for Speech and Environmental Sound Deepfake Detection
2	209	MIF: Multi-view Interaction Framework for Speech and Environment Sound Deepfake Detection
3	215	Dual-Stream EAT-XLSR: A Cascade Framework for Component-Aware Audio Deepfake Detection
4	217	Deepfake Audio Detection Using Self-supervised Fusion Representations
5	221	Phoneme-Guided Fusion for Environment-Aware Speech and Sound Deepfake Detection
6	224	EnvTriCascade: An Environment-Aware Tri-Stage Cascaded Framework for ESDD2 2026 Challenge
7	232	From Signal Separation to Feature Decomposition: A Framework for Synthetic Speech Detection in Complex Environments

4. Panel Discussion

12:00 – 12:20 PM · 20 min

Participants: Xin Wang, Ming Li Host: Xueping Zhang

5. Closing Remarks

12:20 – 12:30 PM · 10 min

Host: Ming Li

Acknowledgement to participants, organizers, reviewers, sponsor, and ICME.

Overall Timeline

Time	Event	Duration	Participants	Host
9:30 – 9:40	Opening Remarks	10 min	—	Ming Li
9:40 – 10:40	Invited Talk & Q&A	60 min	Xin Wang	Ming Li
10:40 – 11:00	Oral Session	20 min	Summary paper	Ming Li
11:00 – 11:10	Coffee Break	10 min	—	Ming Li
11:10 – 12:00	Oral Session & Awards	50 min	All papers	Ming Li
12:00 – 12:20	Panel Discussion	20 min	Ming Li, Xin Wang	Xueping Zhang
12:20 – 12:30	Closing Remarks	10 min	—	Ming Li