Arivox
Developers

A voice-AI API you can actually own

Build agents over a self-hosted ASR + LLM + TTS pipeline — first-class Hebrew, Arabic, Russian and English, sub-second latency, no model lock-in.

Read the API docsGet an API key
POST /v1/calls — cURL
curl https://api.arivox.dev/v1/calls \
  -H "Authorization: Bearer $ARIVOX_KEY" \
  -d '{
    "to": "+972543095860",
    "character": "noa",
    "language": "he",
    "voice": "noa-he",
    "tools": ["lookup_account", "book_appointment"],
    "webhook": "https://acme.co/arivox/events"
  }'
@arivox/sdk — TypeScript
import { Arivox } from "@arivox/sdk";

const client = new Arivox(process.env.ARIVOX_KEY);

const call = await client.calls.create({
  to: "+971501182204",
  character: "layla",
  language: "ar",       // Gulf / Levant / MSA
  voice: "layla-gulf",
  knowledgeBase: "kb_support",
});

console.log(call.id, call.status);

Sub-second latency

A streaming speech-to-speech loop engineered around a ~0.7s voice-to-voice budget.

Region-first engine

language: he | ar | ru | en — first-class, with code-switching, not enterprise add-ons.

Self-host or cloud

Run on your GPUs or ours; point calls at your own deployment endpoint.

Tools, webhooks, memory

Function-calling, mid-call data injection, analysis schemas and persistent memory.

Resources

Everything to ship fast

Quickstart

Make your first call in 5 minutes.

API reference

Full endpoint and SDK reference.

Self-host guide

Deploy on your own GPUs.

Help Center

Product guides and how-tos.

Start building with voice AI for the region.

Grab an API key, or talk to us about a self-hosted deployment.

Get an API keyRead the docs