Introduction
Aura Live is a high-performance, real-time transcription (Automatic Speech Recognition) platform built with FastAPI and designed for low-latency live audio processing.
Core Capabilities
Section titled “Core Capabilities”- Real-time Transcription: Process audio streams and receive text results with minimal latency.
- Multilingual Support: Support for a growing list of languages and dialects.
- WebSocket Protocol: A robust, binary-friendly protocol for streaming audio data.
- REST API: Comprehensive management of languages, models, sessions, and user settings.
- Translation: Real-time translation of transcription output between language pairs.
- Extensible Architecture: Built to integrate with various ASR engine backends.
How it Works
Section titled “How it Works”The typical lifecycle of an Aura Live interaction follows this flow:
- Authentication: The client authenticates using the
/loginendpoint to obtain a JWT token. - Discovery: The client queries the REST API to discover available languages, ASR models, and translation pairs.
- Session Initiation: The client creates a “Live Session” via the REST API.
- Streaming: The client connects to the provided WebSocket URI and streams raw audio data.
- Result Retrieval: Aura Live sends transcription results back through the same WebSocket connection.
Authentication
Section titled “Authentication”Aura Live supports two authentication methods for REST endpoints:
- Bearer Token (
Authorization: Bearer <token>) — JWT obtained from/login. - API Key (
X-API-Key: <key>) — managed from your user settings.
For WebSocket connections, authentication is handled during the initial handshake via a token query parameter.
Target Audience
Section titled “Target Audience”This documentation is intended for:
- Developers looking to integrate real-time transcription into their applications.
- System Integrators connecting Aura Live to existing audio infrastructure.
- Users who want to understand the underlying mechanics of the transcription service.