SDKs

SDKs

Quickstart API reference Error codes Deterministic testing OpenAI compatibility SDKs Changelog

First-party Python and Node packages are in progress and not published yet. Until they are, there are two ways to call the API that work today — and neither needs a dependency.

Nothing to install yet. If you have seen pip install vocalix or npm i vocalix referenced elsewhere, those packages are not on PyPI or npm at the time of writing. Use one of the options below and swap later — the wire format will not change.

1 · The OpenAI SDK, pointed at us

If you already depend on openai, this is the shortest path: change the base URL and keep your code. Full mapping →

2 · Plain HTTP

One POST, one file. There is nothing an SDK would be hiding.

import { writeFile } from "node:fs/promises";

const res = await fetch("https://api.vocalix.ai/v1/tts", {
  method: "POST",
  headers: {
    Authorization: `Bearer ${process.env.VOCALIX_KEY}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    text: "Namaste",
    voice_id: "meera-hi",
    model: "echo-1",
  }),
});

if (!res.ok) throw new Error((await res.json()).code);
await writeFile("reply.wav", Buffer.from(await res.arrayBuffer()));

Error handling is the one place an SDK would earn its keep, and it is three lines: every failure is problem+json with a stable code. The full list →