Reproduce “Fairness of Automatic Speech Recognition: Looking Through a Philosophical Lens” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Anna Seo Gyeong Choi, Hoon Choi

Anna Seo Gyeong Choi and Hoon Choi provide a philosophical framework for understanding Automatic Speech Recognition (ASR) bias, arguing that systematic misrecognition of diverse speech varieties constitutes disrespect and compounds historical injustices, distinct from mere technical limitations. The work identifies unique temporal harms imposed on speakers of non-standard dialects and advocates for ASR system designs that prioritize linguistic pluralism and speaker autonomy.

Key insights

  • ASR bias extends beyond technical limitations, constituting a form of "disrespect" that compounds historical injustices when systematically misrecognizing non-standard speech.
  • The paper uncovers distinct temporal harms specific to ASR, including "temporal taxation" (unequal time burdens), fragmented conversational flow, and inherent power imbalances, often overlooked by current metrics.
  • Systematic misrecognition of speech, integral to identity and autonomy, reinforces problematic standard language ideologies and compels users into "technological code-switching."

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.