Rules

 To support the two challenge tracks, we will release two corresponding datasets: AT-ADD-Track1 and AT-ADD-Track2. 

 Each dataset will include training, validation, and test splits to facilitate the development and evaluation of participants’ countermeasures.

To help participants get started, we will also provide a set of strong baseline systems, including:

• countermeasures based on hand-crafted acoustic features

• countermeasures built on self-supervised learning (SSL) features

• countermeasures based on audio large language model (ALLM) supervised fine-tuning (SFT)

These baselines are intended to provide solid starting points for further research and development.

Further details will be announced in the upcoming official release of the challenge documentation.