Type at the speed of speech.
Hold a hotkey and talk — your words land at the cursor in any Windows app, 4× faster than typing.
RE: Q3 delivery — email
OpenAItext + image
Groqtext
Deepgramtext
ElevenLabstext
Anthropictext
Geminitext + image
Mistraltext
OpenRoutertext + image
Cerebrastext
Ollamatext
Works with any app.
Keep your hands on your coffee. Not your keyboard.
AI enhancement. By app, or on demand.
Say it once. AI rewrites it to fit the moment.
hey um so basically the update is done uh can you check the link and um let tom know we're good for tomorrow
Hi, The update is complete. Could you review the link and confirm with Tom that we're set for tomorrow? Thanks!
your API keys, your provider — Anthropic, OpenAI, Gemini, Groq, Mistral, OpenRouter, Cerebras, Ollama
Names, jargon, product codes — teach it once, spelled right forever.
Every enhancement is an editable prompt. Tweak the built-ins or write your own.
Start or end what you say with a trigger word to pick an enhancement by voice.
Try it on your next email.
Free for 7 days. Then one payment — never a subscription.
Supports 100 languages.
HalloDutch
BonjourFrench
HalloGerman
HolaSpanish
CiaoItalian
OláPortuguese
CześćPolish
HejSwedish
MerhabaTurkish
こんにちはJapanese
안녕하세요Korean
你好Chinese
नमस्तेHindi
ПривітUkrainian
مرحباArabic
Γεια σουGreek
Xin chàoVietnamese
ПриветRussian
HejDanish
HeiNorwegian
MoiFinnish
AhojCzech
SziaHungarian
SalutRomanian
ЗдравейBulgarian
สวัสดีThai
HaloIndonesian
שלוםHebrew
LabasLithuanian
SveikiLatvian
TereEstonian
নমস্কারBengali
HelloEnglish
Voice-to-image — spoken prompt examples
listening…
A slide background, an icon for the spec, a diagram, or something just for the fun of it.
A tour of the app.
Recording, in context.
A floating pill shows what’s recording and where the text will land.
dark theme shown — light and system included
Your voice never has to leave your machine.
No VoiceWink server keeping your history. No account. No voiceprints — your voice is only transcribed. Local by default — cloud only with your own API keys.
Pay once. Keep it.
Solo
Personal
Extended
Try free for 7 days — no credit card.
Windows 11 (x64 & arm64) · GPL v3 · VAT via Lemon Squeezy · Lost your license key?
The optional cloud part? Usually free.
Subscription dictation apps charge $8 to $30 a month for these same engines. VoiceWink is bring-your-own-key: you paste your own API key and requests go straight to the cloud provider, at their prices.
Dictation needs no API key and no account — it runs on the local models. An API key unlocks the extras: dictation on the big cloud models, AI enhancements, and voice-to-image. The free tiers below are the providers' own; for most dictation, they're enough.
Two of them ship, both running on your own CPU. Switch between them any time.
A balance of speed and accuracy for everyday dictation.
Models from small to large — the one to pick when you dictate beyond Parakeet’s languages.
Groq
Whisper on their free tier, no card required. That's more dictation than a voice can physically produce.
Deepgram
A 200-dollar signup credit, no card — more dictation than most people get through in the year it's valid.
ElevenLabs
Scribe transcription on the free plan — renews every month. Enough for light dictation.
OpenAI
No free tier, but pay-as-you-go is cheap — a full hour of transcribed audio costs about 36 cents.
Checked July 2026 — these are the providers' offers, not ours; free tiers change, confirm with each provider · AI Enhancement can ride free tiers too: Groq's covers thousands of requests a day.
Frequently Asked Questions.
Does it work offline?
Yes, by default. Both local engines run on your machine — your audio and transcripts stay on it unless you configure a cloud provider.
Where does my data live?
All of it sits under %LOCALAPPDATA%\VoiceWink, on your machine. Uninstalling leaves it there — you can erase it any time from inside the app, or delete the folder yourself.
What are the requirements?
Windows 11 (x64 build — also runs fine on arm64 PCs), CPU only, no GPU needed. The tiny model runs on any modern laptop; large-v3 is happiest with 8 GB+ of RAM. The installer brings the .NET runtime.
What about Windows 10, macOS, or Linux?
Windows 11 only, at the moment.
Is the installer safe to run?
Yes. Code-signed via Microsoft Trusted Signing (DV Studio), and it updates itself.
Do I need admin rights to install?
No. VoiceWink installs for your user account only, so it needs no admin rights — it works on many locked-down work PCs.
Which local engine should I pick?
Parakeet, unless you need a language it doesn’t cover — then switch to a Whisper model. If you pick a language your model can’t handle, the app says so and suggests one that can.
Does it work in other languages?
Yes — 100 languages via the local Whisper engine (pin one or auto-detect); the faster Parakeet engine covers 25 European languages. Pick a language a model can’t handle and the app says so — and suggests one that can.
Does it paste or type?
It pastes (Ctrl+V) — faster and more compatible than simulated typing.
Does it work with coding agents like Claude Code?
Yes — the text lands wherever your cursor is. If you can type into it, you can talk into it.
Can I write my own AI prompts?
Yes — every enhancement prompt is editable, and you can add your own. Pair them with App Mode and each app picks the right one automatically.
Do I need an API key?
No — local dictation needs no key. VoiceWink works out of the box with local models — your chosen model downloads once during setup: no API key, no account, nothing to sign up for. Keys only matter if you opt into the cloud extras — dictation on the big cloud models, AI enhancements, voice-to-image.
How do I get an API key?
Create a free account with the provider you like and copy the key from their console: OpenAI · Groq · Deepgram · ElevenLabs · Anthropic · Gemini · Mistral · OpenRouter · Cerebras. Then paste it into VoiceWink once — done.
How does bring-your-own-key work?
Paste your API key once — it's encrypted by Windows (DPAPI). Requests go straight to the provider; your account, no middleman.
Can I point it at Ollama or a self-hosted model?
Yes — any OpenAI-compatible base URL works, including local Ollama.
Is there a free trial? What about refunds?
7-day full-featured trial — enter your email at a $0 checkout and your trial key arrives right away. No credit card. After purchase, every tier has a 30-day money-back guarantee.
What happens when the trial ends?
Dictation pauses until you enter a license key. Nothing is deleted — your history, dictionary, and settings are all there when you return.
How does the 30-day money-back guarantee work?
Email support@dvstudio.app within 30 days of purchase with your order number and the email you used at checkout — no reason needed. The refund goes back to your original payment method via Lemon Squeezy, and your license key is deactivated.
Can I move my license to a new PC?
Yes. Deactivate the old machine in the app and activate the new one — tiers count active PCs, not lifetime installs. Old PC dead or wiped? Support frees the slot.
Why GPL v3? What am I paying for?
The full source ships under GPL v3 (a derivative of VoiceInk; the public repo opens at launch, September 2026). Paying gets you signed builds, auto-updates, and support. Self-compiling is free, forever.
More questions? support@dvstudio.app
Start typing at the speed of speech.
Try free for 7 days — no credit card.