paperSeptember 2026Unreviewed

From Scores to Evidence: Auditable Decisions Can Improve Speech Deepfake Detection

Mengzhe Geng, Yujia Lu, Patrick Littell, Manuela Kunz, Xie Chen

Abstract

Speech deepfakes can mimic a speaker's voice convincingly enough to deceive listeners and automated systems. This has driven strong progress in speech deepfake detection, but most detectors still end with one score per utterance. That score is useful for ranking systems, yet it says little about why a borderline item should be trusted, deferred, or reviewed. Two utterances can fall in the same score band for different reasons, for example because passive and retrieval evidence disagree or becaus

Cite

@misc{geng2026from,
  title = {{From Scores to Evidence: Auditable Decisions Can Improve Speech Deepfake Detection}},
  author = {Mengzhe Geng and Yujia Lu and Patrick Littell and Manuela Kunz and Xie Chen},
  year = {2026},
  month = sep,
  eprint = {2609.08899},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2609.08899}
}