← Back to Home
SETUP GUIDE

Getting Started

Vox Bar runs entirely on your machine — no cloud, no API keys, no accounts. Here's everything you need to get set up in under 10 minutes.

Choose Your Setup Path

Vox Bar is designed to fit your hardware. Depending on your OS and engine, setup is either a simple double-click or requires Docker. Find your platform below.

💻

Windows Universal

Free & Lite Tiers

Zero setup required. Works out of the box on almost any Windows PC, including AMD and Intel CPUs.

  • No Docker needed
  • CPU, NVIDIA, or AMD supported
  • 1-2GB Storage
🟩

Windows NVIDIA

Flash, AI, Nemotron, Kyutai

Powerful native models optimized specifically for NVIDIA CUDA. Installs with a single click.

  • No Docker needed
  • Requires NVIDIA GPU
  • 3-8GB Storage depending on tier
🐳

Windows Docker

Pro Tier Only

Our flagship Windows engine requires Docker to orchestrate its advanced multi-modal capabilities.

  • Requires Docker Desktop
  • NVIDIA or AMD (ROCm)
  • 15GB+ Storage
🍎

Apple Silicon Mac

All Tiers Natively

Mac users get a fully native experience. Nemotron, Kyutai, Whisper models, and even Voxtral (Pro) all run entirely locally under Apple Metal. No Docker required.

  • No Docker needed — ever
  • Requires M1/M2/M3/M4
  • Pure native C++ Engine

Detailed Prerequisites

🎮

NVIDIA GPU Drivers

Windows Only

If you plan to use Flash, AI, Nemotron, Kyutai, or Pro with CUDA on Windows, you must have up-to-date NVIDIA drivers. Mac users do not need to worry about driver installation.

⭐ Easiest Method

Download the NVIDIA App (formerly GeForce Experience). It auto-detects your GPU and installs the correct drivers.

Download NVIDIA App
🐳

Docker Desktop

Required for Windows PRO only

The Windows Pro tier runs its advanced Mistral AI model inside a Docker container. Unless you are purchasing Pro on a Windows machine, you can completely skip this step. Mac users never need Docker.

Download Docker Desktop
🔴

AMD GPU Limits

Windows Only

AMD GPUs currently only support the Free (Whisper) and Lite (Moonshine) tiers natively out of the box on Windows.

If you wish to use the Pro tier on an AMD GPU, you must use Docker and set up ROCm via WSL2. The Nemotron, Kyutai, and Canary engines are completely unsupported on AMD at this time.

AMD Driver Download
🖥️

OS & Storage

System Requirements

Vox Bar requires either Windows 10/11 or macOS on Apple Silicon. Because these models run entirely offline, they require actual disk space.

  • Free & Lite: ~1-2 GB
  • Flash & Nemotron: ~3-4 GB
  • AI & Kyutai: ~5-8 GB
  • Pro: ~15 GB+

Quick Start Checklist

Follow these simple steps to start transcribing securely offline.

1

Check Your Hardware & Dependencies

Confirm your GPU aligns with your chosen tier (remember, Nemotron/Kyutai/Flash/AI need NVIDIA on Windows, but they all work natively on Apple Silicon). Install Docker Desktop ONLY if you bought the Pro tier and are running Windows.

2

Purchase & Download Vox Bar

Grab the engine that fits your needs. You'll receive a .zip containing the portable launcher immediately after purchase.

3

Run the Launcher

Extract the file and double click launch.bat or voxbar.app. Everything downloads automatically — no technical configuration needed!

🎉

Start Transcribing!

Click 'Start Listening' and speak. Your voice is securely processed offline.

💡 The Workflow

1.

Overlay — Discreetly see your words appear in front of you as you work. The compact floating bar stays out of your way while you dictate into any application.

2.

Expand to Standard Mode — Need to edit or insert a word? Pop out to the full view where you can review your text, make corrections, and fine-tune before pasting.

3.

Collapse back to Overlay — Done editing? Snap back to the floating bar. It remembers its position on screen and springs right back where you left it.

Overlay Mode for dictation, Standard Mode for editing — switch between them seamlessly.

Have questions? We're here to help.

📧 Contact Support — hello@voxbar.io
Coming Soon