elevenlabs_integration.py

#!/usr/bin/env python3
"""
ElevenLabs + OmniAvatar Integration Example
"""

import requests
import json
import os
from typing import Optional

class ElevenLabsOmniAvatarClient:
    def __init__(self, elevenlabs_api_key: str, omni_avatar_base_url: str = "http://localhost:7860"):
        self.elevenlabs_api_key = elevenlabs_api_key
        self.omni_avatar_base_url = omni_avatar_base_url
        self.elevenlabs_base_url = "https://api.elevenlabs.io/v1"
    
    def text_to_speech_url(self, text: str, voice_id: str, model_id: str = "eleven_monolingual_v1") -> str:
        """
        Generate speech from text using ElevenLabs and return the audio URL
        
        Args:
            text: Text to convert to speech
            voice_id: ElevenLabs voice ID
            model_id: ElevenLabs model ID
            
        Returns:
            URL to the generated audio file
        """
        url = f"{self.elevenlabs_base_url}/text-to-speech/{voice_id}"
        
        headers = {
            "Accept": "audio/mpeg",
            "Content-Type": "application/json",
            "xi-api-key": self.elevenlabs_api_key
        }
        
        data = {
            "text": text,
            "model_id": model_id,
            "voice_settings": {
                "stability": 0.5,
                "similarity_boost": 0.5
            }
        }
        
        # Generate audio
        response = requests.post(url, json=data, headers=headers)
        
        if response.status_code != 200:
            raise Exception(f"ElevenLabs API error: {response.status_code} - {response.text}")
        
        # Save audio to temporary file and return a URL
        # In practice, you might upload this to a CDN or file server
        # For this example, we'll assume you have a way to serve the file
        
        # This is a placeholder - in real implementation, you would:
        # 1. Save the audio file
        # 2. Upload to a file server or CDN
        # 3. Return the public URL
        
        return f"{self.elevenlabs_base_url}/text-to-speech/{voice_id}?text={text}&model_id={model_id}"
    
    def generate_avatar(self, 
                       prompt: str, 
                       speech_text: str, 
                       voice_id: str,
                       image_url: Optional[str] = None,
                       guidance_scale: float = 5.0,
                       audio_scale: float = 3.5,
                       num_steps: int = 30) -> dict:
        """
        Generate avatar video using ElevenLabs audio and OmniAvatar
        
        Args:
            prompt: Description of character behavior
            speech_text: Text to be spoken (sent to ElevenLabs)
            voice_id: ElevenLabs voice ID
            image_url: Optional reference image URL
            guidance_scale: Prompt guidance scale
            audio_scale: Audio guidance scale
            num_steps: Number of inference steps
            
        Returns:
            Generation result with video path and metadata
        """
        
        try:
            # Step 1: Generate audio URL from ElevenLabs
            print(f"🎤 Generating speech with ElevenLabs...")
            print(f"Text: {speech_text}")
            print(f"Voice ID: {voice_id}")
            
            # Get audio URL from ElevenLabs
            elevenlabs_audio_url = self.text_to_speech_url(speech_text, voice_id)
            
            # Step 2: Generate avatar with OmniAvatar
            print(f"🎭 Generating avatar with OmniAvatar...")
            print(f"Prompt: {prompt}")
            
            avatar_data = {
                "prompt": prompt,
                "elevenlabs_audio_url": elevenlabs_audio_url,
                "guidance_scale": guidance_scale,
                "audio_scale": audio_scale,
                "num_steps": num_steps
            }
            
            if image_url:
                avatar_data["image_url"] = image_url
                print(f"Image URL: {image_url}")
            
            response = requests.post(f"{self.omni_avatar_base_url}/generate", json=avatar_data)
            
            if response.status_code != 200:
                raise Exception(f"OmniAvatar API error: {response.status_code} - {response.text}")
            
            result = response.json()
            
            print(f"✅ Avatar generated successfully!")
            print(f"Output: {result['output_path']}")
            print(f"Processing time: {result['processing_time']:.2f}s")
            
            return result
            
        except Exception as e:
            print(f"❌ Error generating avatar: {e}")
            raise

def main():
    """Example usage"""
    
    # Configuration
    ELEVENLABS_API_KEY = os.getenv("ELEVENLABS_API_KEY", "your-elevenlabs-api-key")
    OMNI_AVATAR_URL = os.getenv("OMNI_AVATAR_URL", "http://localhost:7860")
    
    if ELEVENLABS_API_KEY == "your-elevenlabs-api-key":
        print("⚠️ Please set your ELEVENLABS_API_KEY environment variable")
        print("Example: export ELEVENLABS_API_KEY='your-actual-api-key'")
        return
    
    # Initialize client
    client = ElevenLabsOmniAvatarClient(ELEVENLABS_API_KEY, OMNI_AVATAR_URL)
    
    # Example 1: Basic avatar generation
    print("=== Example 1: Basic Avatar Generation ===")
    try:
        result = client.generate_avatar(
            prompt="A friendly teacher explaining a concept with clear hand gestures",
            speech_text="Hello! Today we're going to learn about artificial intelligence and how it works.",
            voice_id="21m00Tcm4TlvDq8ikWAM",  # Replace with your voice ID
            guidance_scale=5.0,
            audio_scale=4.0,
            num_steps=30
        )
        print(f"Video saved to: {result['output_path']}")
    except Exception as e:
        print(f"Example 1 failed: {e}")
    
    # Example 2: Avatar with reference image
    print("\n=== Example 2: Avatar with Reference Image ===")
    try:
        result = client.generate_avatar(
            prompt="A professional presenter speaking confidently to an audience",
            speech_text="Welcome to our presentation on the future of technology.",
            voice_id="21m00Tcm4TlvDq8ikWAM",  # Replace with your voice ID
            image_url="https://example.com/professional-headshot.jpg",  # Replace with actual image
            guidance_scale=5.5,
            audio_scale=3.5,
            num_steps=35
        )
        print(f"Video with reference image saved to: {result['output_path']}")
    except Exception as e:
        print(f"Example 2 failed: {e}")
    
    print("\n🎉 Integration examples completed!")
    print("\nTo use this script:")
    print("1. Set your ElevenLabs API key: export ELEVENLABS_API_KEY='your-key'")
    print("2. Start OmniAvatar API: python app.py")
    print("3. Run this script: python elevenlabs_integration.py")

if __name__ == "__main__":
    main()