Determine whether negative perceptions of ASR captions remain justified

Determine whether continued negative perceptions of automatic speech recognition captions among deaf and hard-of-hearing communities remain grounded in current caption quality or instead reflect inertia despite rapid technological advances.

Background

The study found that ASR captions received viewer ratings similar to mostly human-generated broadcast television captions under synchronized conditions, while several established metrics behaved differently across the two caption sources. These results raise uncertainty about whether prevailing community perceptions of ASR captions accurately reflect their current performance.

The unresolved issue concerns the relationship between technological progress and user perceptions: ASR captions may have become increasingly competitive, but the paper does not establish whether negative perceptions have correspondingly lost their empirical basis.

References

Our findings also raise the question as to whether continued negative perceptions of ASR captions in the DHH communities are still grounded in reality, or whether they are the result of inertia in the face of rapid technological advances.

— Are Caption Metrics Broken? Latency, Deaf and Hard of Hearing User Ratings, and Bias across Technologies  (2609.11408 - Thompson et al., 10 Sep 2026) in Section 6, “Perceptions of ASR vs Reality”