REVIEW 5 cited by
Speech is Silver, Silence is Golden: What do ASVspoof-trained Models Really Learn?
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
Speech is Silver, Silence is Golden: What do ASVspoof-trained Models Really Learn?
read the original abstract
We present our analysis of a significant data artifact in the official 2019/2021 ASVspoof Challenge Dataset. We identify an uneven distribution of silence duration in the training and test splits, which tends to correlate with the target prediction label. Bonafide instances tend to have significantly longer leading and trailing silences than spoofed instances. In this paper, we explore this phenomenon and its impact in depth. We compare several types of models trained on a) only the duration of the leading silence and b) only on the duration of leading and trailing silence. Results show that models trained on only the duration of the leading silence perform particularly well, and achieve up to 85% percent accuracy and an equal error rate (EER) of 15.1%. At the same time, we observe that trimming silence during pre-processing and then training established antispoofing models using signal-based features leads to comparatively worse performance. In that case, EER increases from 3.6% (with silence) to 15.5% (trimmed silence). Our findings suggest that previous work may, in part, have inadvertently learned thespoof/bonafide distinction by relying on the duration of silence as it appears in the official challenge dataset. We discuss the potential consequences that this has for interpreting system scores in the challenge and discuss how the ASV community may further consider this issue.
Forward citations
Cited by 5 Pith papers
-
MixFake: Benchmarking and Enhancing Audio Deepfake Detection in Diverse Real-world Mixed Audio
MixFake is a new benchmark for mixed-authenticity audio and a multi-stream prompt tuning method achieves 0.95% EER foreground and 7.72% absolute gain in complex background deepfake detection.
-
Alethia: A Foundational Encoder for Voice Deepfakes
Alethia is a pretrained audio encoder using continuous embedding prediction and generative flow-matching reconstruction that outperforms existing speech foundation models on voice deepfake tasks with better robustness...
-
DeePen: Penetration Testing for Audio Deepfake Detection
DeePen demonstrates that both production and academic audio deepfake detectors can be reliably deceived by simple signal processing attacks such as time-stretching or echo addition, with some attacks resistible via re...
-
MLAAD: The Multi-Language Audio Anti-Spoofing Dataset
MLAAD provides a large-scale multi-language synthetic audio dataset for training and evaluating audio anti-spoofing models, showing better training performance than InTheWild and FakeOrReal and alternating superiority...
-
Layer-Wise Decision Fusion for Fake Audio Detection Using XLS-R
Per-layer late fusion of one-class softmax classifiers on frozen XLS-R features achieves 6.90% EER on In-the-Wild fake-audio detection, outperforming feature-fusion and single-layer baselines.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.