README

A clean, well-tested Laravel package for the ElevenLabs Voice API.
Supports Text to Speech, Speech to Text, and Sound Effects with a fluent API and Laravel Facade.

Table of Contents

1. Requirements
2. Installation
3. Configuration
4. Usage
   • Text to Speech
   • Speech to Text
   • Sound Effects
   • Listing Voices
5. Available Models
6. Testing
7. Publishing to Packagist
8. Package Architecture Explained

Requirements

• PHP 8.2+
• Laravel 10 or 11
• An ElevenLabs API key

Installation

composer require back2ops/elevenlabs

Laravel's package auto-discovery will automatically register the service provider and ElevenLabs facade — no manual setup needed.

If you have auto-discovery disabled, add these to config/app.php:

'providers' => [
    Back2ops\ElevenLabs\ElevenLabsServiceProvider::class,
],

'aliases' => [
    'ElevenLabs' => Back2ops\ElevenLabs\Facades\ElevenLabs::class,
],

Configuration

1. Publish the config file

php artisan vendor:publish --tag=elevenlabs-config

This creates config/elevenlabs.php in your application.

2. Add your API key to .env

ELEVENLABS_API_KEY=your_api_key_here

# Optional overrides (these have sensible defaults in the config):
ELEVENLABS_VOICE_ID=21m00Tcm4TlvDq8ikWAM   # Rachel (default)
ELEVENLABS_TTS_MODEL=eleven_multilingual_v2
ELEVENLABS_STT_MODEL=scribe_v2
ELEVENLABS_SFX_MODEL=eleven_text_to_sound_v2
ELEVENLABS_OUTPUT_FORMAT=mp3_44100_128
ELEVENLABS_DISK=                            # Leave blank to use default filesystem disk

Usage

You can use the ElevenLabs facade anywhere, or inject Back2ops\ElevenLabs\ElevenLabs directly.

Text to Speech

Get raw audio bytes

use ElevenLabs;

$audioBytes = ElevenLabs::textToSpeech('Hello, welcome to our platform!');

// Store it yourself
Storage::put('audio/welcome.mp3', $audioBytes);

// Or stream it back to the browser
return response($audioBytes, 200)->header('Content-Type', 'audio/mpeg');

With custom options

$audioBytes = ElevenLabs::textToSpeech(
    text: 'Bonjour tout le monde!',
    voiceId: 'EXAVITQu4vr4xnSDxMaL',    // Override voice
    modelId: 'eleven_flash_v2_5',          // Ultra-low latency model
    outputFormat: 'mp3_44100_128',
    voiceSettings: [
        'stability'        => 0.8,
        'similarity_boost' => 0.9,
        'style'            => 0.2,
    ],
    languageCode: 'fr',                    // Force French
);

Generate and store in one call

// Returns the storage path 'audio/greeting.mp3'
$path = ElevenLabs::textToSpeechAndStore(
    text: 'Welcome back!',
    path: 'audio/greeting.mp3',
    disk: 's3',                           // Optional: override filesystem disk
);

// Get the public URL
$url = Storage::disk('s3')->url($path);

Speech to Text

Transcribe a file on disk

// Full response — includes word-level timestamps
$result = ElevenLabs::transcribeFile('/path/to/recording.mp3');

echo $result['text'];           // Full transcript string
echo $result['language_code']; // 'en'

foreach ($result['words'] as $word) {
    echo "{$word['text']} ({$word['start']}s – {$word['end']}s)\n";
}

Just the text

$transcript = ElevenLabs::transcribeToText('/path/to/recording.mp3');
echo $transcript; // "Hello, this is the transcribed content."

From an uploaded file (Controller example)

public function transcribe(Request $request): JsonResponse
{
    $request->validate(['audio' => 'required|file|mimes:mp3,wav,webm,m4a|max:51200']);

    $file = $request->file('audio');

    $result = ElevenLabs::transcribeBytes(
        bytes: file_get_contents($file->getRealPath()),
        filename: $file->getClientOriginalName(),
        language: $request->input('language'),  // e.g. 'en', 'es' — optional
    );

    return response()->json([
        'transcript' => $result['text'],
        'language'   => $result['language_code'],
    ]);
}

Sound Effects

Generate and return bytes

$sfx = ElevenLabs::soundEffect('a heavy wooden door creaking open slowly');

Storage::put('sfx/door-creak.mp3', $sfx);

With options

$sfx = ElevenLabs::soundEffect(
    prompt: 'gentle rain on a tin roof',
    durationSeconds: 10.0,        // 0.5–30 seconds (null = auto-detect)
    promptInfluence: 0.5,         // 0.0–1.0 (default 0.3)
    loopable: true,               // Seamless loop (eleven_text_to_sound_v2 only)
);

Generate and store

$path = ElevenLabs::soundEffectAndStore(
    prompt: 'crowd cheering in a stadium',
    path: 'sfx/crowd-cheer.mp3',
    options: [
        'duration_seconds' => 8.0,
        'loopable'         => false,
    ],
);

Listing Voices

// All voices on your account
$voices = ElevenLabs::voices();

foreach ($voices as $voice) {
    echo $voice['voice_id'] . ' — ' . $voice['name'] . "\n";
}

// Single voice details
$voice = ElevenLabs::voice('21m00Tcm4TlvDq8ikWAM');

// All available models
$models = ElevenLabs::models();

Available Models

Text to Speech

Speech to Text

Sound Effects

Output Formats