Call for Contributions — DCASE AIR Workshop 2026
Applied & Industrial Research (AIR) is a new one-day workshop co-located with DCASE 2026, taking place 27 October 2026 at MIT, Boston, USA (the day before the main DCASE workshop).
DCASE — Detection and Classification of Acoustic Scenes and Events is the leading forum for research on computational audio analysis: acoustic scene classification, sound event detection and localization, audio tagging, anomalous sound detection, audio captioning, and related tasks (for the full technical scope, see the call for papers). While the main workshop focuses on scientific advances and benchmark-driven research, AIR provides a dedicated venue for discussing how DCASE research is being deployed and used in practice.
AIR brings together researchers and practitioners from both academia and industry who are building, deploying, or evaluating real-world audio and acoustic AI systems. If you've taken a DCASE-related method from research into real-world application, or encountered challenges that benchmarks don't capture, this is the venue to share that experience.
Format
Unlike a traditional paper session, AIR is built around technical talks, live demonstrations, and hands-on tutorials. There are no paper submissions. We're looking for contributions that go beyond benchmark performance and share insights into broader systems, engineering decisions, deployment experience, and emerging practical challenges.
Contributions take one of three forms:
- Keynote (~25 min + Q&A) — A technical deep dive into a topic, sharing insights or lessons learned from building real-world systems.
- Demo / Show & Tell (~15 min) — A live on-stage demonstration of a working system, with optional continued demo in a dedicated space afterwards. Hardware, embedded systems, prototypes, and interactive applications all encouraged.
- Tutorial (~25 min + Q&A) — A practical, hands-on introduction to a method, tool, or workflow that attendees can apply in their own work.
Topics of Interest
- Deployed systems and industrial case studies across DCASE tasks
- Engineering challenges: data collection, annotation, and evaluation in the wild
- Embedded and edge audio AI
- Robustness, reliability, and performance beyond benchmarks
- Multimodal and foundation-model applications for audio
- Emerging application areas
- Lessons learned, negative results, and open challenges
We especially encourage contributions that expose engineering trade-offs or identify unsolved problems for the community.
Submission
Please submit your proposed contribution using this link.
Submissions are reviewed on a rolling basis. The call remains open until 1 October 2026 or until all slots are filled.
Registration for AIR is included with DCASE 2026 registration; standalone AIR registration is also available at a discounted rate. Venue and travel details at the DCASE 2026 website.
Organisers
Nicolas Turpault (Sonaid)
Mikolaj Kegler (Bose)
Sile Yin (Bose)
Tornike Karchkhadze (Apple)
Shuo Zhang (Bose)