ElevenLabs TTS CLI

A simple, focused command-line tool for converting text to speech using ElevenLabs.

Features

• Pipe text from stdin or provide as argument
• Voice selection via flags
• Output to file or pipe to audio players
• TypeScript with full type safety
• Configurable voice settings (stability, similarity)

Prerequisites

Option 1: Using Nix Flake with direnv (Recommended)

Automatically sets up the environment when you cd into the directory:

# First time setup
cd elevenlabs/cli
direnv allow

# Environment loads automatically!
# Copy the example env file and add your API key
cp .envrc.local.example .envrc.local
# Edit .envrc.local and add your ELEVENLABS_API_KEY

Or without direnv:

# Enter development shell with all dependencies
nix develop

# Or run directly
nix run . -- "Hello world" -o test.mp3

Option 2: Manual Installation

Requires Bun or Node.js to be installed manually.

Installation

bun install  # or npm install
bun run build  # or npm run build
bun link  # or npm link - Makes 'tts' command available globally

Usage

Basic Examples

# Pipe text and play with mpv
echo "Hello world" | tts | mpv -

# Save to file
echo "Hello world" | tts -o hello.mp3

# Provide text as argument
tts "Hello world" -o hello.mp3

# Use different voice
echo "Hello world" | tts -v "21m00Tcm4TlvDq8ikWAM" | mpv -

Options

-v, --voice <id>         Voice ID (default: ErXwobaYiN019PkySvjV - Antoni)
-o, --output <file>      Output file (default: stdout for piping)
-m, --model <id>         Model ID (default: eleven_multilingual_v2)
-s, --stability <value>  Voice stability 0-1 (default: 0.5)
-S, --similarity <value> Similarity boost 0-1 (default: 0.75)

Common Voice IDs

• ErXwobaYiN019PkySvjV - Antoni (default, male)
• 21m00Tcm4TlvDq8ikWAM - Rachel (female)
• AZnzlk1XvdvUeBnXmlld - Domi (female)
• EXAVITQu4vr4xnSDxMaL - Bella (female)
• MF3mGyEYCl7XYWbV9V6O - Elli (female)
• TxGEqnHWrfWFTfGW9XjX - Josh (male)

Environment Setup

Set your ElevenLabs API key:

export ELEVENLABS_API_KEY="your-api-key-here"

Examples

# Read from file and convert
cat document.txt | tts -o document.mp3

# Generate and play immediately
tts "This is a test" | mpv -

# Use different voice settings
echo "Dramatic reading" | tts -s 0.8 -S 0.9 -o dramatic.mp3

# Pipe from other commands
fortune | tts | mpv -

Development

# Build
npm run build

# Watch mode
npm run dev

# Run directly
npm start -- "Hello world" -o test.mp3