Trade-offs Between Capacity and Robustness in Neural Audio Codecs for Adversarially Robust Speech Recognition
This work addresses adversarial vulnerabilities in speech recognition systems, offering a practical defense mechanism through neural audio codec configuration.
The paper investigates how neural audio codec quantization depth affects adversarial robustness in speech recognition systems, finding that intermediate quantization depths minimize transcription errors by balancing content preservation and perturbation suppression.
Adversarial perturbations exploit vulnerabilities in automatic speech recognition (ASR) systems while preserving human perceived linguistic content. Neural audio codecs impose a discrete bottleneck that can suppress fine-grained signal variations associated with adversarial noise. We examine how the granularity of this bottleneck, controlled by residual vector quantization (RVQ) depth, shapes adversarial robustness. We observe a non-monotonic trade-off under gradient-based attacks: shallow quantization suppresses adversarial perturbations but degrades speech content, while deeper quantization preserves both content and perturbations. Intermediate depths balance these effects and minimize transcription error. We further show that adversarially induced changes in discrete codebook tokens strongly correlate with transcription error. These gains persist under adaptive attacks, where neural codec configurations outperform traditional compression defenses.