Accentus filters speech data: rejects 2% of Americans, 93% of non-American accents
HeyAmit_ · x · 2026-11-06
Luel launched Accentus, a model to detect American English in real-world speech, built because off-the-shelf models either rejected too many real Americans or let too many mismatched recordings through. In testing it rejects under 2% of American speakers while filtering out 93% of tested South Asian, West African, East African, and Filipino speakers, and names accents correctly 96% of the time — a sharp balance point for large-scale speech dataset building, and an ethically contentious one.
More from Models
- FrontierPhysics benchmark: Claude Opus 5.5 tops end-to-end physics research at 27.4% — xeophon · 2026-11-06
- Why AI writing feels annoying despite models winning literary competitions — cis_female · 2026-11-06
- METR chart reveals unreleased Anthropic Model 4 scoring 178.6 on AECI — ResultBackground2450 · 2026-11-06
- Marin's open 535B-A23B MoE finishes main pretraining on 18T tokens, checkpoints soon — dlwh · 2026-11-06
- Artificial Analysis launches Intelligence Index v5 and new Progress Index, reveals eval roadmap — ArtificialAnlys · 2026-11-06
- Fastino open-sources GLiDE: five decision models (0.4B-27B) claiming a new Pareto frontier — HowDevelop · 2026-11-06