DETECT-3B-Omni is Agnostic of Content and Demographics

2026-07-03

AI Summary

To ensure deepfake audio detectors rely solely on acoustic artifacts rather than speech semantics or speaker identity—thereby meeting trustworthiness and GDPR compliance requirements—this study presents the first large-scale, systematic evaluation of the DETECT-3B-Omni detector’s semantic and demographic invariance. The assessment encompasses 10,240 diverse samples generated by eight state-of-the-art AI voice cloning systems. Using equivalence testing at a 99% confidence level, the authors demonstrate that detection accuracy varies by no more than 2 percentage points across any groups defined by content, gender, age, or geographic region. These results confirm the detector’s high robustness to both semantic content and demographic attributes, establishing its suitability for real-world deployment under stringent regulatory and ethical standards.

Abstract

A trustworthy and GDPR-compliant deepfake audio detector must base its decisions on acoustic artifacts, not on what is being said or who is speaking. We present a large-scale study of semantic independence for Resemble AI's detector, DETECT-3B-Omni. Using 10,240 audio samples from diverse US English speakers across 30 states, generated through 8 different AI voice-cloning systems, we test whether detection accuracy depends on spoken content (benign versus malicious), speaker gender, speaker age, or speaker region. Using equivalence testing, our results show that the accuracy difference between any two of these groups is at most 2 percentage points, at 99% confidence. The detector therefore identifies AI-generated audio with equivalent accuracy regardless of what the audio says or who the speaker is.

Problem

• deepfake audio detection
• semantic independence
• demographic bias
• acoustic artifacts
• GDPR compliance

Innovation

• deepfake audio detection
• semantic independence
• acoustic artifacts
• equivalence testing
• GDPR-compliant AI