19 August 2026
13 African languages, at the same rank as French
In most dubbing tools, African languages form a separate box: “other languages,” optional, or missing entirely. At Axeho, they go through the same verification process as French or English. Here is what that actually changes in a studio.
What “same rank” does not mean
It does not mean one more entry in a menu. A language added without being verified helps no one: it gives an adapter detection that gets it wrong, a rythmo band that does not display the right characters, and an empty dictionary the moment they sit down to write. Axeho’s language selector currently lists 26 source languages, 13 of them African. Each one is there because it was measured, not because it fills out a list.
What it actually means
Three things, precisely.
- Accuracy is measured by language, not by continent. Swahili and Hausa sit in the same accuracy tier as Arabic or Korean, not in a separate “African languages” bucket. The ranking follows what was measured, not geography.
- The characters actually render in the band. Yorùbá, Igbo and Wolof use letters that most system fonts do not carry: Ɛ ɛ Ɔ ɔ Ɲ ɲ Ŋ ŋ ẹ ọ ṣ ǹ ñ. A rythmo band that swaps them for a blank box is unusable for these languages, even when the transcription underneath is correct.
- Every source language has its own dictionary. A glossary per language —
wo.mdfor Wolof,yo.mdfor Yorùbá,ln.mdfor Lingala — corrects the AI on proper names and trade expressions, exactly as it does for English or Spanish.
A detail that gives away a system that was measured, not just displayed
Swahili and Hausa, for instance, are not listed under the “African languages” heading in the selector. They sit elsewhere, in the “good accuracy” tier, next to Arabic and Korean — because that is where their measured accuracy actually places them. A system that showed one generic “Africa” box would not have this kind of mismatch between its label and its content, precisely because it never measured much behind the label. Ours does, and keeps it that way: proof that the ranking comes from a measurement, not from a menu heading.
What changes in a studio
An adapter working in Wolof opens the same band, the same dictionary and the same offline lip detection as an adapter working in French. No second, thinner tool, because their language happened to join the list later. That is what “same rank” means: not a stated intention, a consequence you can check inside the software used every day.