BantuNomics My BantuNomics
B BANTUNOMICS
Private access Sign out

The proof point

Frontier ASR trips on the accent.

Mode 3 has many Bantu L1 speakers read the same language-neutral English stories. We transcribed them with a frontier speech-to-text model (google-cloud-stt en-US/latest_long) and scored word-error rate against the known text. The errors are mostly substitutions — the model hears the audio and picks the wrong word. That measurable gap is the asset.

Accent (L1)Takes scoredReference wordsWord error rateWord accuracy
Bemba 20 11575 9.0% 91.0%
Luganda 2 898 8.2% 91.8%

Scored on the consented, de-identified sample. The full eval set — every take, the reference text, and the per-word alignment to reproduce these numbers — ships to Founding Partners.

Hear it

A public sample take.

The Morning at Home · Bemba-accented English · 214.6s · 449 words · ASR WER 9.6%

0–6720 ms The Morning at Home
6720–9240 ms I woke up early.
9240–11280 ms The house was quiet.
11280–13200 ms The sun was not yet high.
13200–15240 ms I opened my eyes slowly.
15240–17520 ms I listened for a moment.
17520–19320 ms I could hear a bird outside.
19320–22200 ms I could hear someone sweeping the yard.

Why it's a clean benchmark

Same text, different mouths.

Because every speaker reads the same neutral English stories, content is held constant and accent is the only variable. That makes the WER comparable across languages and directly usable as a robustness benchmark for any English ASR system.

Coverage matrix Request the eval set