Avatar

Avatars add a visual, human-like presence to your AI agents, creating more engaging and natural interactions. The VideoSDK Agents framework supports virtual avatars through the Simli integration, providing lifelike video representations that sync with your agent's speech.

Overview

Avatar functionality enables your AI agents to:

• Visual Presence: Display a human-like avatar that represents your agent
• Lip Sync: Automatically synchronize avatar mouth movements with speech
• Real-time Rendering: Generate avatar video in real-time during conversations
• Customizable Appearance: Choose from different avatar faces and styles
• Seamless Integration: Works with both CascadingPipeline and RealTimePipeline

What Avatars Enable

With avatar capabilities, your agents can:

• Provide a more human and approachable interface
• Increase user engagement through visual interaction
• Create branded agent personalities with custom appearances
• Enhance accessibility through visual communication cues
• Build trust through consistent visual representation

Simli Avatar Integration

Basic Setup

The Simli avatar integration provides high-quality virtual avatars with real-time lip synchronization.

main.py

from videosdk.plugins.simli import SimliAvatar, SimliConfig  
  
# Configure your avatar 
avatar_config = SimliConfig(  
    apiKey="your-simli-api-key",  
    faceId="0c2b8b04-5274-41f1-a21c-d5c98322efa9",  # Default face  
    syncAudio=True,  
    handleSilence=True  
)  
  
avatar = SimliAvatar(config=avatar_config)

Avatar Configuration Options

Customize your avatar's behavior and appearance:

main.py

from videosdk.plugins.simli import SimliConfig  
  
config = SimliConfig(  
    apiKey="your-simli-api-key",  
    faceId="your-custom-face-id",     # Choose avatar appearance  
    syncAudio=True,                   # Enable lip sync  
    handleSilence=True,               # Manage silent periods  
    maxSessionLength=1800,            # 30 minutes max session  
    maxIdleTime=300                   # 5 minutes idle timeout  
)

Pipeline Integration

Cascading Pipeline

Add avatar to your cascading pipeline setup:

main.py

from videosdk.agents import CascadingPipeline, AgentSession  
from videosdk.plugins.simli import SimliAvatar, SimliConfig  

# Configure avatar
avatar_config = SimliConfig(apiKey="your-simli-api-key")  
avatar = SimliAvatar(config=avatar_config)  

# Create pipeline with avatar
pipeline = CascadingPipeline(  
    stt=your_stt_provider,  
    llm=your_llm_provider,  
    tts=your_tts_provider,  
    avatar=avatar  # Add avatar to pipeline  
)

Real-time Pipeline

Integrate avatar with real-time models:

main.py

from videosdk.agents import RealTimePipeline  
from videosdk.plugins.simli import SimliAvatar, SimliConfig  
from videosdk.plugins.openai import OpenAIRealtime  

# Configure avatar
avatar = SimliAvatar(config=SimliConfig(apiKey="your-api-key"))  

# Configure real-time model
model = OpenAIRealtime(model="gpt-4o-realtime-preview")  

# Create pipeline with avatar
pipeline = RealTimePipeline(  
    model=model,  
    avatar=avatar  
)

info

You can also specify the avatar in your room configuration:

main.py

from videosdk.agents import JobContext, RoomOptions  

def make_context():  
    avatar = SimliAvatar(config=SimliConfig(apiKey="your-api-key"))  
      
    return JobContext(  
        room_options=RoomOptions(  
            room_id="your-room-id",  
            auth_token="your-auth-token",  
            name="Avatar Agent",  
            avatar=avatar  # Specify avatar in room options  
        )  
    )

Example - Try Out Yourself

AI Avatar Example

Complete avatar implementation with weather assistant functionality