Docs / Get started

Introduction

Waridi is an AI research lab training frontier models for human–computer interaction, built around the way people actually speak: regional dialects, accents and code-switching. The voice API is our first product line, and it starts with Swahili.

Why Swahili first? More than 200 million people speak it, and it is the only African language among the official languages of the African Union. It is a bridge language across East and Central Africa, which makes it the right place to begin.

The API lets you turn text into natural speech, turn speech into text, run both in real time, and put a conversational voice agent on a phone line or in a browser.

We build for Tanzanian Swahili first (sw-TZ), with Kenyan Swahili (sw-KE) and Swahili–English code-switching supported from day one. Callers do not speak in one clean language, so neither should your product.

Early access. The API, SDKs and model names on these pages are in preview. Fields may be added, renamed or removed before general availability. We announce every breaking change in the changelog at least 14 days ahead.

What you can build

Product Model / endpoint What it does
Waridi Sauti sauti-standard, sauti-pro · POST /v1/tts Text-to-speech with expressive Swahili voices
Waridi Sikia sikia-1 · POST /v1/stt Speech recognition that handles accents and code-switching
Streaming wss://api.waridi.io/v1/stream Real-time transcription and speech over one WebSocket
Waridi Agents POST /v1/agents Customer-service voice agents that replace press-1 menus
Voice cloning POST /v1/voices/clone Custom voices, only with verified consent of the speaker
Waridi Dub POST /v1/dub Dubbing and localization for film, TV and digital video

Prefer not to write code yet? Waridi Studio, our web app, uses the same models and voices.

How the API is shaped

  • One base URL. Every REST request goes to https://api.waridi.io/v1.
  • JSON in, audio or JSON out. Text-to-speech returns audio bytes. Everything else returns JSON.
  • Bearer keys. Authenticate with an API key in the Authorization header. See Authentication.
  • Swahili by default. If you leave out language, we assume sw-TZ.
  • Request IDs. Every response carries an X-Request-Id header. Include it when you write to us.

Official SDKs

Language Package Install
Python 3.9+ waridi pip install waridi
Node.js 18+ / TypeScript @waridi/sdk npm install @waridi/sdk

Both SDKs are thin wrappers over the REST and WebSocket APIs. Anything you can do with an SDK, you can do with cURL.

Next steps

  1. Get an API key and make your first request in the Quickstart.
  2. Pick a voice in Voices & languages.
  3. For live conversations, read Streaming and Voice agents.

Questions or feedback on the docs: hello@waridi.io.

NextQuickstart