Program

Workshop Program

TrustAudio will take place on Tuesday, November 10, 2026, co-located with AACL-IJCNLP 2026 in Zhuhai, China. The one-day workshop will run from 09:00 to 17:30, with three invited talks, nine oral paper presentations, and the Best Paper Award announcement.

Location: Meeting Room 4 (MR4)

Talk titles, abstracts, and biographies are available on the Speakers page.

Schedule

All times are local to Zhuhai, China (UTC+8).

Time Session Details
09:00–09:10 Opening Remarks Workshop overview and logistics.
09:10–09:50 Invited Talk 1 Shinji Watanabe Groundwork for Open Speech Foundation Models: Data, Recipes, and Task-Specific Models
09:50–10:30 Paper Oral Presentations 1
  1. 09:50–10:10 · 20 min HalluAudio: A Comprehensive Benchmark for Hallucination Detection in Large Audio-Language Models

    Authors: Feiyu Zhao, Yiming Chen, Wenhuan Lu, Daipeng Zhang, Xianghu Yue, Jianguo Wei

  2. 10:10–10:30 · 20 min What Frozen Speech Embeddings Reveal About Audio Deepfake Detection Difficulty

    Authors: Yashas Harshad Patil, Krish Rawat, Anup Shanbhag
10:30–11:00 Coffee Break Networking
11:00–12:00 Paper Oral Presentations 2
  1. 11:00–11:20 · 20 min TeleASR-Amharic: A Benchmark for Robust Amharic ASR in Realistic Telecommunication Conditions

    Authors: Dagnachew Mekonnen Marilign, Enrique Molero, Aabr, Louis Powell, Israel Abebe Azime

  2. 11:20–11:40 · 20 min Transferring Text Jailbreaks to Audio in Tool-Using LLM Agents

    Authors: Murali Emani, Kashyap Manjusha R

  3. 11:40–12:00 · 20 min Activation-Level Characterization of Cross-Model Adversarial Jailbreak Transferability in Speech Language Models

    Authors: Rahaf Ashahrani, Ayana Niwa
12:00–14:00 Lunch Networking
14:00–14:40 Invited Talk 2 Junichi Yamagishi Multi-Layer Defenses Against Speech Deepfakes: Combining Passive Detection and Proactive Protection
14:40–15:30 Paper Oral Presentations 3
  1. 14:40–15:00 · 20 min EXAM2: Extending Audio Understanding in Multilingual and Multimodal

    Authors: Jiawen Wang, Xiaoxue Gao, Zi Haur Pang, Nancy F. Chen

  2. 15:00–15:30 · 30 min Human Perception of Arabic Fake Audio

    Authors: Elsayed Issa, Yi-Fang Cheng
15:30–16:00 Coffee Break Networking
16:00–16:40 Paper Oral Presentations 4
  1. 16:00–16:20 · 20 min Boosting Transferable Adversarial Attacks against Deep Reinforcement Learning

    Authors: Zexin Li, Ruili Yao, Yiming Zeng, Xiaoxue Gao

  2. 16:20–16:40 · 20 min VoiceBank-RAG: Mitigating ASR Errors in Dialect-Rich Telephony Banking Through Agentic Query Normalization

    Author: Olanrewaju Israel Lawal
16:40–17:20 Invited Talk 3 Hung-yi Lee Beyond the Transcript: Safety and Trustworthiness Challenges in Audio-Language Models
17:20–17:30 Closing Remarks & Best Paper Award Announcement Best Paper Award announcement and workshop closing.

Hybrid Attendance

In-person attendance is preferred. Hybrid or remote access details are to be announced if available.