VOX

OPEN SOURCE macOS App Swift 6 v1.4.0 MIT

Voice-Operated eXecution — "Talk to your terminal. Hear what matters."

macOS menu bar app that monitors your development tools and speaks smart summaries. Supports 6 apps (Terminal, iTerm2, Claude Desktop, Cursor, VS Code, Windsurf), auto-detects language, and offers 3 TTS engines including local AI summaries via Ollama. Everything on-device, no cloud, no telemetry.

View on GitHub Download ZIP

Quick workaround: Claude Code Sound Pack

VOX overkill for your setup? This is the lightweight alternative: audio feedback for Claude Code via hooks. Hear when Claude is done, makes an error, or needs your attention — with swappable sound packs (Warcraft Peon, Beavis & Butthead, macOS system sounds, or your own pack). Multiple random variants per event and per-terminal pack setting via env var. No app needed, pure Node.js + hooks.

Read documentation

97%

Swift

1.4.0

Version

Supported Apps

TTS Engines

How it works

VOX monitors your development tools and speaks smart summaries of what happened.

You work

Use Terminal, iTerm2, Claude Desktop, Cursor, VS Code, or Windsurf as usual.

VOX monitors

VOX detects new output via AppleScript and processes it with local AI (Ollama) or heuristics.

VOX speaks

A smart summary is spoken in the detected language using your chosen TTS engine.

VOX never executes commands — it only reads terminal output. No accessibility permissions, no shell access, no risk.

Features

Compact, powerful, and fully on-device.

Multi-App Support

Monitors Terminal.app, iTerm2, Claude Desktop, Cursor, VS Code, and Windsurf. Automatic background detection.

Adaptive Voices

Auto-detects Dutch, English, and German content. Speaks summaries in the matching language with appropriate voice.

3 TTS Engines

macOS native (NSSpeechSynthesizer), Edge TTS (Microsoft voices), and ElevenLabs. Choose quality vs. speed.

Local AI Summaries

Ollama-powered on-device summarization. No cloud required. Falls back to heuristic summaries when Ollama is unavailable.

Sound Pack Store

MyInstants integration for downloadable sound effects. Custom event sounds for completions, errors, and alerts.

Command History

Searchable, filterable log of all monitored commands and responses. Persistently stored as JSON.

Verbosity Levels

Choose how much detail you want to hear. The Summary level uses smart heuristics: git status shows file counts, build output detects success/failure, errors automatically increase verbosity.

Level	Name	What you hear
0	Silent	Nothing (visual indicator only)
1	Ping	"Done." or "Error occurred."
2	Summary	Heuristic 1-2 sentence summary (default)
3	Full	Complete response read aloud

Privacy & Security

VOX is built with privacy as its foundation. Everything runs locally on your Mac.

No Cloud

All processing happens on-device. No server, no API calls.

No Telemetry

Zero analytics, zero tracking, zero data collection.

No Audio Storage

Voice audio is never stored. Hex handles all input.

No Shell Access

VOX reads output via AppleScript, never executes commands.

No Permissions

No accessibility or microphone permissions required.

Open Source

Full source code is public and auditable.

Tech Stack

Native macOS app built with modern Swift.

Language

Swift 6 with SwiftUI for the UI

Text-to-Speech

macOS native, Edge TTS, ElevenLabs

AI Summaries

Ollama (local, on-device) + heuristic fallback

Terminal Reading

AppleScript via /usr/bin/osascript

Platform

macOS 14+ (Apple Silicon optimized)

Testing

XCTest unit tests

Requirements

macOS 14.0 (Sonoma) or later

Ollama Optional, for AI summaries

Xcode 16+ / Swift 6 (to build)

Permissions None required

Installation

Build from source with Swift Package Manager.

# Clone the repository
git clone https://github.com/RichardTheuws/VOX-app.git
cd VOX-app

# Build and install to /Applications
./scripts/build-app.sh --install

# Or build, install and launch
./scripts/build-app.sh --install --open

# Debug build
swift build

# Run tests
swift test
            

Project Structure

Overview of the app architecture.

VOX/ ├── Assets/ │ ├── AppIcon.icns App icoon │ └── Info.plist App configuratie ├── Sources/VOX/ │ ├── VOXApp.swift @main entry point met MenuBarExtra │ ├── AppState.swift Centrale coordinator (Hex → Monitor → TTS) │ ├── Models/ │ │ ├── VerbosityLevel.swift 4-level verbosity enum │ │ ├── TargetApp.swift Terminal, iTerm2, Claude Code, VS Code │ │ ├── VoxCommand.swift Command history model │ │ └── VoxSettings.swift @AppStorage settings │ ├── Services/ │ │ ├── HexBridge.swift Monitort Hex transcripties │ │ ├── TerminalReader.swift Leest terminal via AppleScript │ │ ├── ResponseProcessor.swift Verbosity-aware samenvattingen │ │ ├── TTSEngine.swift Text-to-speech engine │ │ └── CommandHistory.swift Persistent command log │ └── Views/ │ ├── MenuBarView.swift Menu bar dropdown │ ├── SettingsView.swift 4-tab instellingen │ ├── HistoryView.swift Doorzoekbare history │ └── OnboardingView.swift 3-staps wizard ├── Tests/VOXTests/ 32 unit tests ├── scripts/ Build scripts ├── Package.swift Swift Package Manager └── LICENSE MIT

Roadmap

From MVP to multi-app voice companion.

v0.1–0.3 Terminal monitoring + macOS Say TTS + floating panels + onboarding DONE

v0.4–0.6 Hex file monitoring + monitor mode + push-to-talk companion DONE

v1.0 Multi-app support (6 apps) + Edge TTS + command history DONE

v1.2 ElevenLabs TTS + adaptive voice detection (NL/EN/DE) DONE

v1.4 Ollama local AI summaries + Sound Pack Store + verbosity levels CURRENT

v2.0 Plugin system + custom app adapters + Homebrew install PLANNED

Contributing

VOX is open source and welcomes contributions from everyone.

View on GitHub Report Issue

License: MIT · Tags: voice-control, macos, swift, developer-tools, speech-to-text