Package Exports

@okw/stt
@okw/stt/dist/index.js

This package does not declare an exports field, so the exports above have been automatically detected and optimized by JSPM instead. If any package subpath is missing, it is recommended to post an issue to the original package (@okw/stt) to support the "exports" field. If that is not possible, create a JSPM override to customize the exports field for this package.

Readme

`stt` - Text to Speech CLI 🎤

Give your AI agent a voice using ElevenLabs API. Streams audio directly to your terminal.

Why?

AI coding agents can use this CLI to speak to you - announce completions, read errors aloud, or provide audio feedback while you work.

Requirements

Node.js 18+
mpv for audio playback
ElevenLabs API key (get one here)

# macOS
brew install mpv

Setup

Set your ElevenLabs API key:

export ELEVENLABS_API_KEY=your_api_key_here

Usage

# Basic usage (recommended)
npx @okw/stt "Hello, world!"

# Save to file instead of playing
npx @okw/stt -o output.mp3 "Hello, world!"

# Use a different voice
npx @okw/stt -v CwhRBWXzGAHq8TQ4Fs17 "Hello, world!"

# Adjust speech speed
npx @okw/stt --speed 1.5 "Faster speech"

# Use audio tags (v3 model)
npx @okw/stt "I'm so excited! [laughs] This is amazing!"

Global Installation

If you use it frequently, install globally:

npm install -g @okw/stt
stt "Hello, world!"

Give Your AI Agent a Voice

Add this to your AGENTS.md to let your AI agent speak:

## Voice Output

You can speak to the user using: `npx @okw/stt "Your message"`

Use sparingly for:
- Task completion announcements
- Critical errors
- When user explicitly asks for voice output

Keep messages to 1-2 sentences. Don't read everything aloud.

Options

Option	Description	Default
`-v, --voice <id>`	Voice ID	`UgBBYS2sOqTuMpoF3BR0`
`-m, --model <id>`	Model ID	`eleven_v3`
`-s, --stability <n>`	Voice stability (0-1)	`0.5`
`-b, --boost <n>`	Similarity boost (0-1)	`0.75`
`--speed <n>`	Speech speed (0.5-2.0)	`1.0`
`-o, --output <file>`	Save to file instead of playing	-

Available Voices

Run this to list available voices:

curl -s "https://api.elevenlabs.io/v1/voices" \
  -H "xi-api-key: $ELEVENLABS_API_KEY" | \
  jq -r '.voices[] | "\(.voice_id) | \(.name)"'

Popular premade voices:

ID	Name
`CwhRBWXzGAHq8TQ4Fs17`	Roger - Laid-Back, Casual
`EXAVITQu4vr4xnSDxMaL`	Sarah - Mature, Reassuring
`IKne3meq5aSn9XLyUdCD`	Charlie - Deep, Confident
`onwK4e9ZLuTAKqWW03F9`	Daniel - Steady Broadcaster
`pFZP5JQG7iQjIQuC4Bku`	Lily - Velvety Actress

Models

Model	Description
`eleven_v3`	Most expressive, supports audio tags
`eleven_turbo_v2_5`	Low latency, good quality
`eleven_flash_v2_5`	Ultra-low latency (<75ms)
`eleven_multilingual_v2`	Best for non-English

License

MIT

@okw/stt