Skip to main content

​
Introduction

The LangShazam API provides real-time spoken language detection through a WebSocket-based service. Send audio data and receive language identification results powered by OpenAI’s Whisper model.

​
Base URL

The production API is deployed on AWS Kubernetes: 
wss://3.149.10.154.nip.io/ws
The frontend automatically connects to this endpoint via the ServerDiscovery service. See Server Configuration for alternative deployment endpoints.

​
Authentication

No authentication is required for the WebSocket endpoints. The service uses an internal OpenAI API key for processing audio data.

​
Core Endpoints

WebSocket Connection

Real-time audio streaming endpoint at /ws

Metrics

Server health and performance metrics at /metrics

​
Quick Start

const ws = new WebSocket('wss://3.149.10.154.nip.io/ws');

ws.onopen = () => {
  // Send audio data as binary
  ws.send(audioBuffer);
};

ws.onmessage = (event) => {
  const response = JSON.parse(event.data);
  console.log('Detected language:', response.data.language);
};

​
Response Format

All responses follow a standardized JSON format:
​
status
string
required
Response status: success or error
​
data
object
Result data when status is success
​
message
string
Error message when status is error
​
timestamp
string
required
ISO 8601 timestamp of the response
​
connection_id
string
required
Unique identifier for the connection (8-character UUID)

​
Rate Limits

The service supports a maximum of 3 concurrent OpenAI API calls. Additional requests are queued automatically.

​
CORS Configuration

The following origins are allowed:
  • https://www.langshazam.com
  • https://langshazam.com
  • http://localhost:3000
  • http://localhost:5173
  • http://127.0.0.1:3000
  • http://127.0.0.1:5173

Build docs developers (and LLMs) love

Get started for freeTalk to us