Skip to main content
Welcome to the AIVOCO Development Guide — your reference for building, testing, and scaling speech-to-speech (S2S) voice agents powered by the ROSE model.

​
Local Development Setup

Before you start coding, ensure that you’ve completed the Quickstart guide. You’ll need:

​
WebSocket Connection

The ROSE model communicates in real time over WebSocket. To connect your application: 
const ws = new WebSocket("wss://call.aivoco.on.cloud.vispark.in/ws/{api_key}/{agent_id}");

ws.onopen = () => {
  console.log("Connected to ROSE agent");
  // Start sending or receiving audio data here
};

ws.onmessage = (event) => {
  console.log("Received audio or transcript:", event.data);
};

ws.onclose = () => console.log("Connection closed");
This WebSocket stream enables real-time bi-directional audio between your system and the ROSE model.

​
Agent Lifecycle

You can manage your voice agents through the REST API. Common operations include:
ActionEndpoint
Create AgentPOST /agents
Get AgentGET /agents/{id}
Update AgentPUT /agents/{id}
Delete AgentDELETE /agents/{id}
Each operation allows you to programmatically manage agent configuration, personalities, voices, and function calling capabilities.

​
Connecting with Telephony

To use your agent for voice calls, connect the WebSocket endpoint to your telephony provider. We provide an example integration with Twilio in Outbound Telephony Integration, but you can use any telephony system that supports audio streaming. If your preferred provider is not listed, you can reach out at 📧 [email protected] for custom integration support.

​
Testing and Debugging

You can test your agents using the ROSE WebSocket without deploying to production.
Simply connect using your API key and agent ID from the playground.
Record incoming and outgoing audio locally to verify audio stream quality.
Use tools like ffmpeg or sox to capture and inspect audio packets.
The WebSocket will send both audio and transcript events.
Log all message events in your client for better observability.

​
Retrieving Transcriptions

After your calls are completed, you can retrieve full conversation transcripts using the GET /transcriptions/{call_id} endpoint. This allows you to analyze and store the conversation context for your business logic or analytics layer.

​
Next Steps

  • Build and deploy your own ROSE voice agent
  • Connect with your telephony or VoIP system
  • Retrieve and analyze transcripts
  • Launch your AI-powered conversations
© 2025 AIVOCO | ROSE — Real-Time Speech-to-Speech Intelligence