# ElevenLabs Eleven v4 - Core.Today AI API

> ElevenLabs Eleven v4 text-to-speech via Fal.AI. Expressive speech with inline audio tags ([whispering], [excited]), IPA pronunciation, stability/similarity controls, up to 5,000 characters per request and optional character-level timestamps.

- **Provider**: ElevenLabs
- **Model ID**: elevenlabs/tts/eleven-v4
- **Category**: Audio & TTS
- **Credits**: 186 per 1000 characters (0.186 credits/char, billed per character)
- **Speed**: Medium
- **Quality**: Ultra

## Features
- Expressive delivery steered by inline audio tags like [whispering] and [excited]
- IPA pronunciation control enclosed in forward slashes
- Stability and similarity controls to shape delivery
- Up to 5,000 characters per request with optional character-level timestamps

## Use Cases
- Narration and voiceover for videos, ads and explainers
- Audiobook and long-form reading with emotional nuance
- Character dialogue for games and interactive content
- Subtitle or lip-sync alignment using character-level timestamps

## 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 - The text to convert to speech (max 5,000 characters, billed per character). Supports audio tags such as [whispering] or [excited] and IPA pronunciation enclosed in forward slashes.

### Optional
- **voice**: string (default: Rachel) - The voice to use — an ElevenLabs premade voice name (e.g. Rachel, Aria, Roger, Sarah, George) or a voice ID.
- **stability**: number (default: 0.5) - Voice stability (0-1). Lower values allow more expressive delivery; higher values make delivery more consistent.
- **similarity_boost**: number (default: 0.75) - How closely the output follows the reference voice (0-1). Higher values increase similarity but may reduce naturalness.
- **language_code**: string - Language code (ISO 639-1, e.g. en, ko, ja) for speech generation and text normalization. Auto-detected when omitted.
- **apply_text_normalization**: string (default: auto) - Whether to normalize text such as numbers and dates before generation.
  Options: auto, on, off
- **output_format**: string (default: mp3_44100_128) - Output audio format, formatted as codec_sample_rate_bitrate (MP3 variants).
  Options: mp3_22050_32, mp3_44100_32, mp3_44100_64, mp3_44100_96, mp3_44100_128, mp3_44100_192
- **timestamps**: boolean (default: false) - Whether to return character-level timing information with the generated audio.
- **seed**: integer - Seed for best-effort reproducibility. Identical output is not guaranteed.

## Examples

### Excited greeting
A short line using an audio tag to set an excited delivery.
```json
{
  "model": "elevenlabs/tts/eleven-v4",
  "input": {
    "text": "[excited] Hello! Welcome to Eleven v4.",
    "voice": "Aria"
  }
}
```

### Whispered Korean narration
Korean narration with a whispered delivery and an explicit language code.
```json
{
  "model": "elevenlabs/tts/eleven-v4",
  "input": {
    "text": "[whispering] 조용히 해 봐요. 숲이 잠들고 있어요.",
    "voice": "Rachel",
    "language_code": "ko",
    "stability": 0.4
  }
}
```

## 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
- Billing counts every character of text, including audio tags — keep tags purposeful.
- Lower stability for more expressive, varied delivery; raise it for consistent narration.
- Set language_code when the text mixes languages or when auto-detection picks the wrong one.
- Use Eleven v4 Turbo for the same controls at half the per-character rate.

## Documentation
https://fal.ai/models/elevenlabs/tts/eleven-v4