Neural Codec Source Tracing: Toward Comprehensive Attribution in Open-Set Condition

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Xie, Yuankun, Wang, Xiaopeng, Wang, Zhiyong, Fu, Ruibo, Wen, Zhengqi, Cao, Songjun, Ma, Long, Li, Chenxing, Cheng, Haonnan, Ye, Long
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866915099052081152
author Xie, Yuankun
Wang, Xiaopeng
Wang, Zhiyong
Fu, Ruibo
Wen, Zhengqi
Cao, Songjun
Ma, Long
Li, Chenxing
Cheng, Haonnan
Ye, Long
author_facet Xie, Yuankun
Wang, Xiaopeng
Wang, Zhiyong
Fu, Ruibo
Wen, Zhengqi
Cao, Songjun
Ma, Long
Li, Chenxing
Cheng, Haonnan
Ye, Long
contents Current research in audio deepfake detection is gradually transitioning from binary classification to multi-class tasks, referred as audio deepfake source tracing task. However, existing studies on source tracing consider only closed-set scenarios and have not considered the challenges posed by open-set conditions. In this paper, we define the Neural Codec Source Tracing (NCST) task, which is capable of performing open-set neural codec classification and interpretable ALM detection. Specifically, we constructed the ST-Codecfake dataset for the NCST task, which includes bilingual audio samples generated by 11 state-of-the-art neural codec methods and ALM-based out-ofdistribution (OOD) test samples. Furthermore, we establish a comprehensive source tracing benchmark to assess NCST models in open-set conditions. The experimental results reveal that although the NCST models perform well in in-distribution (ID) classification and OOD detection, they lack robustness in classifying unseen real audio. The ST-codecfake dataset and code are available.
format Preprint
id arxiv_https___arxiv_org_abs_2501_06514
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Neural Codec Source Tracing: Toward Comprehensive Attribution in Open-Set Condition
Xie, Yuankun
Wang, Xiaopeng
Wang, Zhiyong
Fu, Ruibo
Wen, Zhengqi
Cao, Songjun
Ma, Long
Li, Chenxing
Cheng, Haonnan
Ye, Long
Sound
Artificial Intelligence
Audio and Speech Processing
Current research in audio deepfake detection is gradually transitioning from binary classification to multi-class tasks, referred as audio deepfake source tracing task. However, existing studies on source tracing consider only closed-set scenarios and have not considered the challenges posed by open-set conditions. In this paper, we define the Neural Codec Source Tracing (NCST) task, which is capable of performing open-set neural codec classification and interpretable ALM detection. Specifically, we constructed the ST-Codecfake dataset for the NCST task, which includes bilingual audio samples generated by 11 state-of-the-art neural codec methods and ALM-based out-ofdistribution (OOD) test samples. Furthermore, we establish a comprehensive source tracing benchmark to assess NCST models in open-set conditions. The experimental results reveal that although the NCST models perform well in in-distribution (ID) classification and OOD detection, they lack robustness in classifying unseen real audio. The ST-codecfake dataset and code are available.
title Neural Codec Source Tracing: Toward Comprehensive Attribution in Open-Set Condition
topic Sound
Artificial Intelligence
Audio and Speech Processing
url https://arxiv.org/abs/2501.06514