The Open ASR Leaderboard has added two new private evaluation splits for English accents, in collaboration with Liva AI: Egyptian English (1.34 hours, conversational, spontaneous) and Nigerian English (1.55 hours, conversational, spontaneous). Both are transcribed with punctuation, casing, and disfluencies. This follows earlier private data additions from Appen and DataOceanAI covering five accents, and a later Voice Arena split for Indian-accented English. The new splits are part of a continued effort to improve accent and language representation in ASR evaluations, and to strengthen trust by providing new held-out private data that minimizes benchmaxxing.

With the addition, the leaderboard’s Private tab can now show a richer accent breakdown based on Kachru’s Three Circles, an idea credited to Hume AI’s leaderboard. The reported groupings are US vs. non-US (UK, Australia, Canada, India, Nigeria, Egypt), and Inner Circle (US, UK, Australia, Canada) vs. Outer & Expanding Circles (India, Nigeria, Egypt). As more accents are added, the Outer and Expanding Circles will be split further.

Liva AI, the data provider, collects multimodal interaction data through consumer products used by real people. Because it owns both the collection environment and distribution, it can design products around specific model gaps across domains, industries, languages, modalities, and interaction types. This gives it access to natural, high-context behavior that emerges through real product use, rather than artificially elicited or task-based interactions.