Text-to-speech V2: 24 English accents and 20 Arabic dialects
The second generation of our own text-to-speech is live on every workspace. More accents, more dialects, and a voice catalog of 433 voices that an agent can be switched to in one click.
A voice agent is judged in the first three seconds. If it sounds like it belongs on the line, the caller settles. If it sounds like it was recorded somewhere else, the caller starts looking for the zero key. The second generation of our text-to-speech is built for that first impression: it speaks the way people in each market actually speak, and it does so on a streaming path fast enough for a live phone call.
V2 ships today on every Kejue workspace. Existing agents keep the voice they have. New voices are one pick away on the agent's Config tab.
Arabic, by dialect and by region
The catalog now covers 20 Arabic dialects, from Modern Standard Arabic to Gulf, Levantine, Egyptian, Iraqi and the Maghreb, and inside the larger ones regional voices as well. A Saudi agent can be Najdi, Hejazi, Qassimi or Southern. An Egyptian one can be Cairo, Alexandria or Saidi.
Dialect is a property of the voice, not a prompt instruction. Pick the voice and the agent speaks that way on every turn, with pronunciation tables for your names, numbers and brands applied on top.
- 390 Arabic voices across 20 dialects
- Regional voices inside the larger dialects, each tied to its parent
- Saudi 70 voices, Egyptian 58, Jordanian 37, Emirati 27, Syrian 25
English in 24 accents
English grows from 20 accents to 24. Ten are native accents, including American, British, Australian, Irish, Indian, South African and Nigerian, with British split further into Southern, Northern, Scottish, Welsh and Northern Irish. Fourteen more are English as it is spoken by Arabic, French, Spanish, Turkish, Persian and other first-language speakers, which is how a great many customers in this region actually sound on the phone, and how an agent serving them can sound too.
The voice picker. Search by name, dialect or language; each voice carries its origin.
One picker, every voice
All 433 voices sit in one list on the agent's Config tab. Search by name, dialect or language, hear a sample, pick it. The same picker drives the pronunciation editor, so you can hear a corrected name in the exact voice that will say it on the call.
Speaking rate is adjustable per voice. Everything else about a voice, its gender, dialect and register, is in the catalog and shown on the chip next to the name, so nothing has to be approximated in the prompt.
What else
shipped.
Product · 20 October 2026Speech-to-text V2: faster, more accurate, and better with everyday speech
The second generation of our own speech recognition is live. It is faster, it is more accurate, and it understands colloquial speech and slang across the dialects it was trained on.
Product · 6 October 2026Groups: A/B test agents on live calls
Put two or more agents in a group, point a phone number or a campaign at it, and compare them on real traffic. Weighted, sticky per caller, measured on every call, and honest about when the numbers are not in yet.