Teffic-Audio: Tell Fact from Fiction

2026-07-30

AI Summary

This work addresses the generalization challenge in deepfake speech detection caused by heterogeneous attacks—such as synthesis, conversion, and codec processing—and varying environmental factors. To this end, we propose a lightweight yet effective universal detection system featuring a streamlined architecture composed of a Conformer encoder, multi-head attention-based statistical pooling, and a binary classifier. Without increasing model complexity, our approach leverages only publicly available data and enhances generalization through multi-source attack and speaker-balanced sampling alongside diverse audio augmentation strategies. Evaluated on 14 test sets from Speech-DF-Arena, the system achieves an average equal error rate (EER) of 1.454%, ranking first among all public systems, and attains the lowest EER on five subsets, demonstrating both superior performance and practical utility.

Abstract

Speech deepfake detection has expanded in scope with increasingly heterogeneous spoofing mechanisms, including speech synthesis, voice conversion, vocoder reconstruction, and neural-codec resynthesis. The resulting spoofing artifacts can be further shaped by variability in source speech, recording environments, and transmission channels. This variability makes robust generalization across heterogeneous conditions a central requirement for practical detection systems. This report presents Teffic-Audio, a general speech deepfake detection system designed for comprehensive evaluation environment. Teffic-Audio adopts a straightforward detector architecture consisting of a Conformer-based speech encoder, multi-head attentive statistics pooling, and a binary classifier. Rather than relying on additional architectural complexity, the system improves generalization through its training recipe, which integrates multi-source data, attack- and source-balanced sampling, and diverse audio augmentation. Trained only with open-source data, Teffic-Audio achieves a pooled EER of 1.454% on the 14 test sets of Speech-DF-Arena, outperforming all currently public systems on the leaderboard. It also obtains the lowest EER on five individual test sets and shows a favorable performance-complexity trade-off compared with larger leading systems. Overall, Teffic-Audio provides a strong and practical reference system for general speech deepfake detection.

Problem

• speech deepfake detection
• spoofing mechanisms
• robust generalization
• heterogeneous conditions
• audio forensics

Innovation

• speech deepfake detection
• generalization
• Conformer-based encoder
• balanced sampling
• audio augmentation