Audio codec improves deepfake detection on low bitrate recordings
Neural Audio Codec for Robust Audio Deepfake Detection
SoundMultimedia
Summary
Audio deepfake detectors usually work on high-quality audio, but real-world clips are often compressed, which makes detecting fake sounds harder. The authors studied how different compression methods affect detection and found that low-quality codecs can cause errors. To fix this, they created a special neural audio codec that keeps important clues for detecting fakes, even after compression. This new codec greatly lowers errors without losing sound quality and works well with different fake audio detectors.
What this means in practice
- •For audio security teams: Detect fake audio more reliably in compressed recordings encountered in real-world security and authentication setups.
- •For streaming platform engineers: Integrate a codec that preserves deepfake detection clues to improve content verification on low-bitrate audio streams.
Authors
Jungwoo Kim, Joonyong Park, Junyoung Koh, Jong-Seok Lee
Abstract
Audio deepfake detectors are typically evaluated on uncompressed audio, although real-world audio often undergoes low-bitrate coding. In this work, we investigate how audio coding affects deepfake detection across codecs, bitrates, and detectors, finding higher errors at lower rates. A mixed-pair protocol isolates codec-induced changes in bona fide and spoof audio, revealing asymmetric, codec-dependent failures: low-rate DAC and EnCodec mainly degrade bona fide detection, whereas X-Codec shows a stronger spoof-side limitation. Motivated by these, we propose a forensic-preserving neural audio codec (FP-NAC), which fine-tunes a pretrained codec using a detector-guided objective while preserving its native hard quantization path and bitrate. On ASVspoof 2019 LA, FP-NAC reduces EER by up to 49.8~pp compared with the original DAC at 0.5~kbps while maintaining comparable reconstruction quality. Although supervised by only one detector, FP-NAC improves performance across multiple detectors, highlighting forensic transparency as a codec design objective alongside perceptual quality. Our codes are available at https://github.com/kjungwoo03/FP-NAC.