How Navoras works

Navoras turns text into speech in Kenyan English and Kiswahili, in voices recorded by Kenyan voice actors under contract. You send text to the API, or type it in the dashboard, and get back audio, with the timing of every word if you ask for it.

#What happens to your text

1. 1

   Your text is read the Kenyan way

   Before a voice sees your text, Navoras writes out what a model would misread: shillings, phone numbers, paybill and till numbers, dates, times (in Swahili time when the sentence is in Kiswahili) and the names in our lexicon.

   How Navoras reads Kenyan text
2. 2

   Each sentence is spoken in its own language

   A sentence in Kiswahili is read in Kiswahili, one in English in English, so text that mixes the two sounds natural. You can also set the language for the whole request.

3. 3

   You get audio and timings back

   MP3, WAV, PCM, Opus and the telephone formats. The timestamps endpoints add the start and end of every character, both of your text as written and of what was said.

#How it's paid for

Speech costs credits: one credit a character of text on the standard model, the same in Kiswahili as in English. A plan grants credits every month, and you can buy more whenever you need them. Asking for the same words again with the same seed costs nothing, and test keys are always free.