Are Modern Speech Enhancement Systems Vulnerable to Adversarial Attacks?
Fuente:
arXiv
Salvato in:
| Autori principali: | , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866909002276798464 |
|---|---|
| author | Makarov, Rostislav Schönherr, Lea Gerkmann, Timo |
| author_facet | Makarov, Rostislav Schönherr, Lea Gerkmann, Timo |
| contents | Machine learning approaches for speech enhancement are becoming increasingly expressive, enabling ever more powerful modifications of input signals. In this paper, we demonstrate that this expressiveness introduces a vulnerability: advanced speech enhancement models can be susceptible to adversarial attacks. Specifically, we show that adversarial noise, carefully crafted and psychoacoustically masked by the original input, can be injected such that the enhanced speech output conveys an entirely different semantic meaning. We experimentally verify that contemporary predictive speech enhancement models can indeed be manipulated in this way. Furthermore, we highlight that diffusion models with stochastic samplers exhibit inherent robustness to such adversarial attacks by design. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_21087 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Are Modern Speech Enhancement Systems Vulnerable to Adversarial Attacks? Makarov, Rostislav Schönherr, Lea Gerkmann, Timo Audio and Speech Processing Machine Learning Sound Machine learning approaches for speech enhancement are becoming increasingly expressive, enabling ever more powerful modifications of input signals. In this paper, we demonstrate that this expressiveness introduces a vulnerability: advanced speech enhancement models can be susceptible to adversarial attacks. Specifically, we show that adversarial noise, carefully crafted and psychoacoustically masked by the original input, can be injected such that the enhanced speech output conveys an entirely different semantic meaning. We experimentally verify that contemporary predictive speech enhancement models can indeed be manipulated in this way. Furthermore, we highlight that diffusion models with stochastic samplers exhibit inherent robustness to such adversarial attacks by design. |
| title | Are Modern Speech Enhancement Systems Vulnerable to Adversarial Attacks? |
| topic | Audio and Speech Processing Machine Learning Sound |
| url | https://arxiv.org/abs/2509.21087 |