Overview of Addis AI API

The Addis AI API provides integration capabilities for Ethiopian language technology in applications. The platform offers services for Amharic and Afan Oromo.

Introduction to Addis AI

Addis AI is a language technology provider focused on Ethiopian languages. The API provides functionality for:
  • Conversational AI: Text-based chatbots and assistants
  • Text-to-Speech: Converting text to spoken audio
  • Speech-to-Text: Transcribing spoken audio to text
  • Multi-modal Processing: Handling text, audio, and image inputs
The goal is to provide AI capabilities for languages that have limited support in existing language technology platforms.

Supported Languages

Addis AI currently supports two Ethiopian languages:
  • Amharic (am):
    • The official working language of the Federal Democratic Republic of Ethiopia
    • Uses the Ge'ez script
    • Language code: am
  • Afan Oromo (om):
    • A Cushitic language spoken in Ethiopia and the Horn of Africa
    • Uses a Latin-based alphabet (Qubee Afaan Oromoo)
    • Language code: om

API Structure

The Addis AI API follows RESTful design with JSON-based requests and responses.

Technical Design:

  • Request/Response Format: Structured JSON for standard requests, multipart/form-data for file uploads
  • Error Handling: Standardized error codes and messages across all endpoints
  • Performance: Optimized for standard request-response patterns
  • Multi-modal Support: API accepts text, audio files, and image attachments
  • Stateless Architecture: No server-side state management; conversation context can be maintained client-side through the conversation history parameter
  • Security: HTTPS-only communication with API key authentication

Core Components:

  1. Authentication: API key verification system
  2. Endpoints:
    • /chat_generate - Text generation and chat functionality
    • /audio - Text-to-speech conversion
  3. Language Models: Specialized models for Amharic and Afan Oromo
  4. Input Processing: Handles text input and file uploads
  5. Response Generation: Produces formatted text and audio responses
The API is designed to provide reliable Ethiopian language capabilities with a focus on accuracy and functionality rather than speed or real-time processing.