[2603.03359] ACES: Accent Subspaces for Coupling, Explanations, and Stress-Testing in Automatic Speech Recognition

Computer Science > Sound arXiv:2603.03359 (cs) [Submitted on 28 Feb 2026] Title:ACES: Accent Subspaces for Coupling, Explanations, and Stress-Testing in Automatic Speech Recognition Authors:Swapnil Parekh View a PDF of the paper titled ACES: Accent Subspaces for Coupling, Explanations, and Stress-Testing in Automatic Speech Recognition, by Swapnil Parekh View PDF HTML (experimental) Abstract:ASR systems exhibit persistent performance disparities across accents, yet the internal mechanisms underlying these gaps remain poorly understood. We introduce ACES, a representation-centric audit that extracts accent-discriminative subspaces and uses them to probe model fragility and disparity. Analyzing Wav2Vec2-base with five English accents, we find that accent information concentrates in a low-dimensional early-layer subspace (layer 3, k=8). Projection magnitude correlates with per-utterance WER (r=0.26), and crucially, subspace-constrained perturbations yield stronger coupling between representation shift and degradation (r=0.32) than random-subspace controls (r=0.15). Finally, linear attenuation of this subspace however does not reduce disparity and slightly worsens it. Our findings suggest that accent-relevant features are deeply entangled with recognition-critical cues, positioning accent subspaces as vital diagnostic tools rather than simple "erasure" levers for fairness. Subjects: Sound (cs.SD); Artificial Intelligence (cs.AI); Audio and Speech Processing (eess.AS) Cite as: ...