​

Replicate

Integration with Replicate AI platform for running various AI models including image generation, audio synthesis, and other machine learning operations.

​

Overview

The Replicate skill provides functionality for:

• Image generation and manipulation
• Text-to-speech and audio processing
• Background removal and image editing
• Running various AI models on demand
• Managing file outputs with configurable expiration

​

Connection Requirements

This skill uses an internal service and doesn’t require external connections.

​

Basic Usage

// Basic image generation
POST: {INTERNAL_SKILL_URL}/replicate/run
{
  "authorId": "black-forest-labs",
  "modelId": "flux-schnell",
  "input": {
    "prompt": "A serene mountain landscape at sunset",
    "aspect_ratio": "16:9"
  }
}

​

Key Features

​

Image Generation

• Fast Generation: Quick image creation with flux-schnell
• High Quality: Detailed images with flux-dev
• Custom Aspects: Various aspect ratios and formats
• Multiple Outputs: Generate multiple variations

​

Audio Processing

• Text-to-Speech: Convert text to natural speech
• Voice Selection: Multiple voice options and emotions
• Speed Control: Adjustable speech rate
• Audio Formats: Various output formats

​

Image Editing

• Background Removal: Automatic background removal
• Image Transformation: Various image processing operations
• Format Conversion: Convert between image formats
• Quality Control: Adjustable output quality

​

Supported Content Types

​

✅ Supported

• Images: Generate, edit, and transform images
• Audio: Generate speech and audio processing

​

❌ Not Supported

• Video: Video generation and processing not available

​

Common Operations

​

Image Generation (Fast)

POST: replicate/run
{
  "authorId": "black-forest-labs",
  "modelId": "flux-schnell",
  "input": {
    "prompt": "A futuristic cityscape at night",
    "aspect_ratio": "16:9",
    "num_outputs": 1,
    "output_format": "webp",
    "go_fast": true
  }
}

​

Image Generation (Quality)

POST: replicate/run
{
  "authorId": "black-forest-labs",
  "modelId": "flux-dev",
  "input": {
    "prompt": "Portrait of a person in Renaissance style",
    "aspect_ratio": "1:1",
    "num_outputs": 1,
    "output_format": "png"
  }
}

​

Text-to-Speech

POST: replicate/run
{
  "authorId": "minimax",
  "modelId": "speech-02-turbo",
  "input": {
    "text": "Welcome to our application. How can I help you today?",
    "voice_id": "Young_Woman",
    "speed": 1.0,
    "emotion": "friendly"
  }
}

​

Background Removal

POST: replicate/run
{
  "authorId": "851-labs",
  "modelId": "background-remover",
  "input": {
    "image": "https://example.com/image.jpg",
    "format": "png"
  }
}

​

Recommended Models

​

Image Generation

• flux-schnell: Fast image generation (black-forest-labs/flux-schnell)
• flux-dev: High-quality image generation (black-forest-labs/flux-dev)

​

Audio Processing

• speech-02-turbo: Text-to-speech conversion (minimax/speech-02-turbo)

​

Image Editing

• background-remover: Automatic background removal (851-labs/background-remover)

​

Request Structure

​

Required Parameters

{
  "authorId": "string",     // Model author (e.g., "black-forest-labs")
  "modelId": "string",      // Model identifier (e.g., "flux-schnell")
  "input": {}              // Model-specific input parameters
}

​

Optional Parameters

{
  "version": "string",              // Specific model version
  "fileLinksExpireInDays": 7        // File expiration (1-30 days)
}

​

Response Structure

​

Successful Response

{
  "output": [
    {
      "url": "https://replicate.delivery/...",
      "mimeType": "image/webp"
    }
  ]
}

​

Audio Response

{
  "output": [
    {
      "url": "https://replicate.delivery/...",
      "mimeType": "audio/wav"
    }
  ]
}

​

Model Parameters

​

Flux Image Models (flux-schnell/flux-dev)

• prompt: Text description (required)
• aspect_ratio: “1:1”, “16:9”, “9:16”, “4:3”, “3:2”
• num_outputs: Number of images (1-4)
• output_format: “webp”, “jpg”, “png”
• go_fast: Boolean for speed optimization (flux-schnell only)

​

Speech Model (speech-02-turbo)

• text: Text to convert (required)
• voice_id: “Deep_Voice_Man”, “Young_Woman”, “Mature_Man”
• speed: Speech rate (0.5-2.0)
• emotion: “auto”, “happy”, “friendly”, “sad”, “angry”

​

Background Remover

• image: Image URL (required)
• format: Output format (“png”, “jpg”)

​

File Management

​

File Expiration

• Default: 7 days
• Range: 1-30 days
• Configurable per request

​

File URLs

• Files are hosted on Replicate’s CDN
• URLs are temporary based on expiration setting
• Download files if long-term storage needed

​

Aspect Ratios

​

Common Ratios

• 1:1: Square format
• 16:9: Widescreen landscape
• 9:16: Portrait/mobile format
• 4:3: Traditional landscape
• 3:2: Photography standard

​

Voice Options

​

Available Voices

• Deep_Voice_Man: Deep male voice
• Young_Woman: Young female voice
• Mature_Man: Mature male voice
• Professional_Woman: Professional female voice

​

Emotions

• auto: Automatic emotion detection
• neutral: Neutral tone
• happy: Cheerful tone
• friendly: Warm, approachable tone
• sad: Melancholic tone
• angry: Intense tone

​

Important Notes

• Content Types: Only images and audio are supported (no video)
• File Expiration: Plan for file expiration when using generated content
• Model Versions: Some models may have specific version requirements
• Rate Limits: Respect usage limits for high-volume operations
• Input Validation: Validate input parameters based on model requirements

​

Best Practices

1. Model Selection: Use flux-schnell for speed, flux-dev for quality
2. Prompt Engineering: Write clear, descriptive prompts for better results
3. File Management: Download important files before expiration
4. Error Handling: Implement retry logic for failed generations
5. Cost Optimization: Choose appropriate models for your use case
6. Content Guidelines: Follow Replicate’s content policy guidelines
7. Version Pinning: Use specific model versions for consistency
8. Batch Processing: Group similar requests for efficiency