Orvyn — Real-Time Conversational Avatar Platform (API-First)

Speak Face-to-Face with AI Avatars in Real-Time

Orvyn streams lip-synced digital human avatars over WebSocket & MSE. Integrate ultra-low latency voice dialogue and fluid neural video into your native products.

✦ Real-time active listening & natural speech response

✦ The avatar hears your audio input naturally

✦ Zero client-side LLM or TTS complexity — authoritative worker stream

How Orvyn Works

Orvyn decouples voice intelligence, neural avatar rendering, and web delivery into a single unified API pipeline.

STEP 01 // AUDIO INGRESS

Speak Naturally

The visitor speaks into their microphone. Audio is streamed over WebSocket directly to the Orvyn Worker session bridge.

STEP 02 // INTELLIGENCE ROUTING

Voice & Response Routing

Orvyn routes user speech through configured voice engines and LLM providers to generate intelligent dialogue responses.

STEP 03 // REAL-TIME RENDER

fMP4 Avatar Delivery

The neural GPU renderer returns lip-synced avatar frames at 25 FPS, streamed back seamlessly to the browser MediaSource.

Designed for High-Touch Experiences

Replace static forms and text bots with lifelike avatar interaction across your user journey.

Customer Support

Empathetic, real-time avatar support agents that resolve complex inquiries with active listening and instant visual rapport.

Integrates via Orvyn JS SDK

Onboarding & Education

Interactive avatar instructors that guide users through platform walk-throughs, answering questions dynamically in natural voice.

Dynamic Avatar Persona Switching

Sales & Product Guidance

Convert prospects with face-to-face avatar sales representatives tailored to demonstrate product features on-demand.

Low Latency Sub-400ms Turnaround

Developer First

A Simple REST & WebSocket API

Initiate avatar sessions with a single API call, then stream media over standard WebSocket connections.

✦ Short-lived HMAC client tokens for zero key exposure
✦ Standard fMP4 over MSE video playback
✦ Full event notifications for turn starts, audio deltas, and completion

POST /v1/sessions cURL / JSON

curl -X POST https://api.orvyn.live/v1/sessions \
  -H "Authorization: Bearer avtr_live_..." \
  -H "Content-Type: application/json" \
  -d '{
    "provider_type": "elevenlabs",
    "avatar_config": {
      "avatar_id": "arnold",
      "bg_id": "bluegradient"
    },
    "transport": "websocket_fmp4"
  }'

// Response:
{
  "session_id": "sess_89f2a412",
  "client_token": "hmac_tok_90a1b2...",
  "transport": "websocket_fmp4"
}