How It Works
A few guided steps, from a voice in your head to clean, finished audio.
Pick or clone a voice
Pick a ready-made voice, blend two for something unique, or drop a short sample to clone a real one.
Train it in one click
Vox runs the whole training pipeline for you — prep, align, train — with a live progress bar and a running timer (it takes a few minutes). No notebooks, no command line.
Generate anything
Type your text and Vox speaks it in that voice. Export a clean WAV, ready to use.
Pick a voice, or blend your own
Choose from ready-made voices and hear each one instantly. Want something unique? Blend two into a voice that’s nobody else’s — no recording session required.
- Pick from ready-made voices, hear them instantly
- Blend two voices into one that’s yours
- Or clone a real voice from a short clip
Create the voice
Trained in one click
Vox handles the whole training pipeline for you — prep, align, train — and shows a live progress bar the whole way. No Python, no notebooks, no command line.
- One-click auto-training, start to finish
- Live, step-by-step progress
- Each voice is self-contained & reusable
Train the voice
One click teaches the voice using your clips — keep this page open until it finishes.
100% Local. 100% Yours.
Everything runs on your computer. No API key, no per-use fees, no uploads — your voices and audio never leave it. Generate as much as you want at $0 per run; a voice SaaS bills you per minute, every month, for the same work.
- Unlimited generation at $0 per run — forever
- No API key, nothing to pay per use
- Generates offline — nothing uploaded
- Licensed so you can sell what you make
File manager
Everything for the selected voice — and exactly where it lives on your computer.
Made for people who need a voice
If you need a voice on demand and don’t want a subscription or a recording session, Vox is for you.
Voiceover & narration
A consistent narrator voice on tap for your scripts.
Video creators
Voice your shorts and explainers without ever picking up a mic.
Audiobooks & e-learning
Narrate books, articles or course scripts — one voice, no per-word fees (long books run in a few passes).
Game & app devs
Generate character lines and UI prompts locally, royalty-free.
Podcasters
Re-voice, dub or patch lines in a voice that stays consistent.
Why local & one-time
You buy it once and run it on your own GPU. Compare that to a per-minute voice SaaS.
Inside the app
An illustrative look at the Vox Studio workflow.
Make the voice talk
Create the voice
Add practice clips
Paste a link or upload audio — long files are fine, we cut them into clips.
Train the voice
One click teaches the voice using your clips — keep this page open until it finishes.
File manager
Everything for the selected voice — and exactly where it lives on your computer.
Everything Included
One powerful desktop app. No add-ons. No hidden costs.
Pick
Choose from ready-made voices, then blend two for something unique.
Clone
Learn a custom voice from your own clips.
Train
One-click custom training with live progress.
Generate
Type text, get speech. Export a clean WAV.
Voice library
Unlimited voices, each self-contained and reusable.
Fully local
No API, nothing to pay per use, nothing uploaded.
Technical Specs
What it runs on and what’s inside — all local, no accounts.
Windows 10 / 11 desktop app
NVIDIA GPU required, 8 GB+ VRAM
Generates offline — no account, no API
Pick: a library of ready-made voices, blendable
Clone: train a custom voice from your own clips
Train: your own custom voice model
Export clean WAV, ready to drop in
Build an unlimited library of voices
Each voice self-contained — copy, back up, reuse
English voice output (v1)
Pricing
Buy once. No subscription, ever.
- Every feature — unlimited voices and generation
- Commercial rights: sell what you make
- Free updates for this version
- Activate on up to 2 computers
- No subscription, no per-use fees, no API key
Less than a few months of a voice subscription — and then it’s yours.
⚠ Before you buy: Vox needs an NVIDIA GPU (8 GB+ VRAM) and about 40 GB free disk. With no NVIDIA card, Vox will not run. See full requirements.
Buy Vox Studio or pay with crypto →Requirements
Windows 10 / 11 desktop app.
NVIDIA GPU required — 8 GB VRAM minimum, 10 GB+ recommended. Vox runs its AI on your own card and will not start without an NVIDIA GPU.
About 40 GB free disk for first-time setup (about 32 GB after). The first run downloads roughly 21 GB of voice models.
Designing or cloning a voice fetches one more model (about 2 GB) the first time — a one-off download.
Generates English (v1).
Runs offline once set up — only the first-time setup goes online, to download the models.
A clean voice sample if you want to clone one (use voices you have the rights to).
One license = up to 2 computers.
Buy once. Updates included, no recurring fee.
What's New
Buy once and every update to this version is included.
- Pick from ready-made voices and blend two into one
- Voice cloning that learns from your own clips
- One-click custom training
- Generate & export WAV — fully local, no API
Questions
Is this a subscription?
No — Vox Studio is a one-time purchase. You own this version forever and get its updates. No monthly fee, ever.
Does it cost anything to run?
No. Everything runs on your own computer — no API key, no per-minute or per-character fees, nothing uploaded. Once you own it, generating is free.
Do I need a GPU?
Yes — an NVIDIA GPU is required. Vox runs its AI on your own machine and will not start without an NVIDIA card (8 GB+ VRAM, with an up-to-date driver). There’s no cloud fallback — that’s the trade for unlimited, private, no-subscription generation.
Can I clone a real voice?
Yes — give Vox a few minutes of clean audio of the voice, and it learns a custom voice from your clips (a short prep, then a quick training run). Only use voices you have the rights to.
Can I sell what I make?
Yes. The audio you generate is yours to use commercially. Vox is built on open-source voice engines, each under its own license — full credits are in the app’s open-source notices.
How many voices can I make?
As many as you like. Each voice is self-contained in its own folder — easy to copy, back up and reuse across projects.
What languages does it support?
Vox v1 generates English. You can design, blend, or clone any English-speaking voice and have it read English text. Support for more languages may come in a future version.
What does it run on?
A Windows 10 / 11 desktop app with an NVIDIA GPU; a macOS version is in development. One license activates on up to 2 computers.
Does my data leave my computer?
No. Your samples, voices, and generated audio all stay on your computer — we never receive them. Vox only goes online to download the voice models on first run, check for updates, verify your license, and — only if you paste a link — fetch that clip from its source.
What about refunds?
Vox needs an NVIDIA GPU — please check the requirements above before buying. If it’s still not the right fit, see the refund policy for details.
See It in Action
A quick look at Vox Studio in under two minutes.
From the blog
Go deeper on Vox Studio
Guides, breakdowns, and what’s new — the blog keeps growing, so there’s always more to dig into.