# An Empirical Study on Channel Effects for Synthetic Voice Spoofing Countermeasure Systems

arXiv: https://arxiv.org/abs/2104.01320 · HTML: https://mlsanity.com/paper/2104.01320/

## Key facts

- Cited by 10 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 3 Apr 2021.

## Built on by (most cited first, 10 of 10)

- [SingFake: Singing Voice Deepfake Detection](https://mlsanity.com/paper/2309.07525/) (2023)
- [SAMO: Speaker Attractor Multi-Center One-Class Learning for Voice Anti-Spoofing](https://mlsanity.com/paper/2211.02718/) (2022)
- [What to Remember: Self-Adaptive Continual Learning for Audio Deepfake Detection](https://mlsanity.com/paper/2312.09651/) (2023)
- [Do You Remember? Overcoming Catastrophic Forgetting for Fake Audio Detection](https://mlsanity.com/paper/2308.03300/) (2023)
- [Multi-Dataset Co-Training with Sharpness-Aware Optimization for Audio Anti-spoofing](https://mlsanity.com/paper/2305.19953/) (2023)
- [Adaptive Fake Audio Detection with Low-Rank Model Squeezing](https://mlsanity.com/paper/2306.04956/) (2023)
- [Dual-Branch Knowledge Distillation for Noise-Robust Synthetic Speech Detection](https://mlsanity.com/paper/2310.08869/) (2023)
- [Deep Spectro-temporal Artifacts for Detecting Synthesized Speech](https://mlsanity.com/paper/2210.05254/) (2022)
- [FADEL: Uncertainty-aware Fake Audio Detection with Evidential Deep Learning](https://mlsanity.com/paper/2504.15663/) (2025)
- [Impact of Channel Variation on One-Class Learning for Spoof Detection](https://mlsanity.com/paper/2109.14900/) (2021)

## Abstract (arXiv)

Spoofing countermeasure (CM) systems are critical in speaker verification; they aim to discern spoofing attacks from bona fide speech trials. In practice, however, acoustic condition variability in speech utterances may significantly degrade the performance of CM systems. In this paper, we conduct a cross-dataset study on several state-of-the-art CM systems and observe significant performance degradation compared with their single-dataset performance. Observing differences of average magnitude spectra of bona fide utterances across the datasets, we hypothesize that channel mismatch among these datasets is one important reason. We then verify it by demonstrating a similar degradation of CM systems trained on original but evaluated on channel-shifted data. Finally, we propose several channel robust strategies (data augmentation, multi-task learning, adversarial learning) for CM systems, and observe a significant performance improvement on cross-dataset experiments.