# Chatterbox Turbo - Core.Today AI API

> Resemble AI's fastest open-source TTS without sacrificing quality. 20 pre-made voices, paralinguistic tags like [sigh] and [chuckle], and optional instant voice cloning from 5s+ reference audio. Max 500 characters per request.

- **Provider**: Resemble AI
- **Model ID**: resemble-ai/chatterbox-turbo
- **Category**: Audio & TTS
- **Credits**: 59 per 1000 characters (0.057 credits/char, billed per character)
- **Speed**: Fast
- **Quality**: High

## Features
- Fastest open-source TTS quality tier
- Paralinguistic tags: [sigh], [chuckle], [cough], [clear throat]
- 20 pre-made voices
- Instant voice cloning from 5s+ reference audio
- Very low per-character price

## Use Cases
- Real-time conversational agents
- Cost-efficient bulk narration
- Natural-sounding customer service bots
- Personalized voice messages
- Prototype voice interfaces

## API Endpoint

Base URL: https://api.core.today/v1
Create Prediction: POST /predictions
Get Status: GET /predictions/{job_id}

## Authentication
Header: X-API-Key: YOUR_API_KEY

## Input Parameters

### Required
- **text**: string - Text to synthesize (max 500 characters). Supports tags like [sigh], [chuckle], [cough]

### Optional
- **voice**: string (default: Andy) - Pre-made voice. Ignored if reference_audio is provided
  Options: Aaron, Abigail, Anaya, Andy, Archer, Brian, Chloe, Dylan, Emmanuel, Ethan, Evelyn, Gavin, Gordon, Ivan, Laura, Lucy, Madison, Marisol, Meera, Walter
- **reference_audio**: string - Reference audio for voice cloning (must be longer than 5 seconds). Overrides voice
- **temperature**: number (default: 0.8) - Controls randomness (0.05-2.0). Higher produces more varied speech
- **top_p**: number (default: 0.95) - Nucleus sampling threshold (0.5-1.0). Lower is more focused
- **top_k**: integer (default: 1000) - Limits vocabulary to top k tokens (1-2000)
- **repetition_penalty**: number (default: 1.2) - Penalizes token repetition (1.0-2.0)
- **seed**: integer - Random seed for reproducible results. Leave blank for random

## Examples

### Natural Conversational Reply
Customer-service style reply with paralinguistic tags
```json
{
  "model": "resemble-ai/chatterbox-turbo",
  "input": {
    "text": "Oh, that's hilarious! [chuckle] Anyway, we do have a new model in store. Would you like me to get some prices for you?",
    "voice": "Laura",
    "temperature": 0.8
  }
}
```

## Response Format
```json
{
  "job_id": "abc123",
  "status": "pending | processing | completed | failed",
  "result": "URL or data (when completed)"
}
```

## Usage Flow
1. POST /predictions with model and input -> receive job_id
2. GET /predictions/{job_id} -> poll until status is completed or failed
3. Result contains output URL(s)

## Tips
- Keep requests under 500 characters — split longer scripts into segments
- Use tags like [sigh] and [chuckle] sparingly for natural conversational feel
- Provide 10-20s of clean reference_audio for the best voice clone quality
- Set a fixed seed to reproduce a delivery you like

## Documentation
https://replicate.com/resemble-ai/chatterbox-turbo