voxclaw

VoxClaw

Give OpenClaw a voice.

OpenClaw is the open-source personal AI assistant that runs on your devices — your files, your shell, your messaging apps (WhatsApp, Telegram, Slack, Discord, and more). It lives where you work. VoxClaw gives it a voice.

Run VoxClaw on your Mac and hear OpenClaw speak to you. When OpenClaw runs on another computer — a server, a headless box, or a different machine — send text to your Mac over the network and VoxClaw speaks it aloud with high-quality text-to-speech. Apple's built-in voices work out of the box; add your own OpenAI or ElevenLabs API key for neural voices when you want that extra polish. Paste text, pipe from the CLI, or stream from any device on your LAN — and listen.

VoxClaw also includes an iPhone app in this repo (VoxClawIOS/) with the same core listener + teleprompter flow for iOS.

For Codex plugin installation and marketplace distribution, use the separate malpern/codex-marketplace repository. This repository is the main product repo for the app, website assets, releases, and source code.

---

A macOS menu bar app + CLI tool that reads text aloud using Apple TTS (default), OpenAI TTS (BYOK), or ElevenLabs TTS (BYOK), with an optional teleprompter-style floating overlay and synchronized word highlighting.

Screenshots

Overlay panel

Settings panel

Features

• Onboarding + Agent Handoff — First-run setup configures voice, key, network mode, and launch at login, and can copy a 🦞 VoxClaw setup pointer for your agent
• Teleprompter Overlay — Floating overlay with word-by-word highlighting synced to speech; includes presets, deep appearance controls, and audio-only mode
• Three Voice Engines — Apple (no setup), OpenAI (BYOK), or ElevenLabs (BYOK), with Apple fallback when cloud auth fails
• Multiple Input Methods — Arguments, stdin pipe, file, clipboard, URL scheme, and LAN HTTP
• Network API for Agents — POST /read, GET /status, and GET /claw, with request validation and structured status payloads
• Bonjour Discovery — Advertises _voxclaw._tcp on LAN for peer/device discovery
• Menu Bar App + CLI — Lightweight menu bar controls plus full terminal control via voxclaw
• iPhone Companion App — iOS app target included in this repo (VoxClawIOS/) using the shared core engine/settings stack
• macOS Services Integration — Read selected text from other apps via Services
• Keyboard Controls While Reading — Space (pause/resume), Escape (stop), Arrow keys (skip ±3s)

Installation

Prerequisites

• macOS 26+
• OpenAI API key (optional)
• ElevenLabs API key (optional)

The onboarding wizard walks you through setup on first launch. To store an API key manually:

security add-generic-password -a "openai" -s "openai-voice-api-key" -w "sk-..."

Or set the environment variable:

export OPENAI_API_KEY="sk-..."

Build & Install

swift build -c release
./Scripts/package_app.sh
./Scripts/install-cli.sh

iPhone App (iOS)

The repo includes an iPhone app target.

Open /Users/malpern/local-code/VoxClaw/VoxClawIOS/VoxClawIOS.xcodeproj in Xcode and run the VoxClawIOS scheme on iOS 26+.

Usage

CLI

voxclaw "Hello, this is a test."       # direct text
echo "Read this aloud" | voxclaw       # piped stdin
voxclaw --file ~/speech.txt            # from file
voxclaw --clipboard                    # from clipboard
voxclaw --audio-only "No overlay"      # audio only, no panel
voxclaw --voice nova "Hello"           # OpenAI voice override
voxclaw --rate 1.5 "Hello"            # 1.5x speech speed
voxclaw --output hello.mp3 "Hello"    # save audio to file (OpenAI)
voxclaw --listen                       # network mode: listen for text from LAN
voxclaw --send "Hello from CLI"        # send text to a running listener
voxclaw --status                       # check if listener is running
voxclaw                                # launch menu bar app (no args)

Network Mode

Start VoxClaw in network listener mode on your Mac: