The authors applied three post-training compression methods (Wanda pruning, HQQ and NF4 quantization, and distillation) to the Whisper family (tiny to large-v3) and evaluated demographic fairness across three datasets: FairSpeech, Common Voice 25, and AfriSpeech-200.
Compression Widens Demographic Fairness Gaps in Whisper Speech Recognition Models
Pruning more than doubles the correction time gap between Black/AA and Asian speakers; quantization causes catastrophic loops on West African accents; distillation narrows gaps in most settings.
Academic
Srishti Ginjala · Eric Fosler-Lussier · Christopher W. Myers · Srinivasan Parthasarathy
The Ohio State University · Air Force Research Laboratory
Research Digest··3 min read
Thread:Compression Fairness
Ginjala et al.
Why this paper
From The Ohio State University and Air Force Research Laboratory
In one line
Post-training pruning and quantization can sharply amplify demographic error burdens in Whisper, while distillation usually narrows them.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ·No stated limitations found
- ✓Reports numbers on named benchmarks
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§