A new open-set neural codec source tracing benchmark and dataset shows strong in-distribution classification and OOD detection, but poor generalization to unseen real audio.
Baseline model This section introduces the baseline models used for the NCST task, including three representative models in the field of ADD: Mel-LCNN, AASIST, and W2V2-AASIST
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SD 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Neural Codec Source Tracing: Toward Comprehensive Attribution in Open-Set Condition
A new open-set neural codec source tracing benchmark and dataset shows strong in-distribution classification and OOD detection, but poor generalization to unseen real audio.