Automatic speech recognition (ASR) systems have achieved high accuracy with transformer-based models, enabling deployment in critical applications. However, they remain vulnerable to adversarial manipulation, particularly in black-box settings where attacks must preserve perceptual naturalness. This work introduces GATAS, a black-box testing approach that generates failure inducing inputs by operating in the phoneme-level latent space of a text- to-speech model. Instead of perturbing waveforms directly, the approach interpolates latent representations to induce transcription errors while remaining within the manifold of natural speech. The attack is formulated as a multi-objective optimization problem balancing semantic divergence and perceptual quality. Our empirical evaluation against both white-box and black-box baselines shows that GATAS achieves a 98% success rate while producing lower distortion and higher perceptual quality, as confirmed by human studies. Despite operating without gradient access, GATAS remains competitive against white-box methods, highlighting that representation and perceptual alignment are more critical than access to model internals. Overall, our results demonstrate that untargeted latent-space optimization enables the efficient generation of realistic and effective test cases for ASR systems.
Generative Testing of Automated Speech Recognition Systems
Automatic speech recognition (ASR) systems have achieved high accuracy with transformer-based models, enabling deployment in critical applications. However, they remain vulnerable to adversarial manipulation, particularly in black-box settings where attacks must preserve…
- Preview

- Year
- 2026
- Hosting
- Full text hostedCC-BY-4.0
Cite
Notes
Only stored in your browser.
Attribution
- Abstract & full text
- arxiv.org/abs/2607.09833CC-BY-4.0
- TL;DR
- Semantic Scholar