Browser-native voice to text

Speak here.
Write anywhere.

Turn rough speech into ready-to-send text without leaving the browser field you are working in. Choose on-device, your own AI key, or a managed plan—and see exactly where your words are processed.

Choose your plan See how it works

Local firstAudio stays on your device by default
54 languagesTranslate directly as you dictate
$49 onceOwn the full BYOK version

ToProduct teamSubjectThursday launch review

Message

Hi team,

Let's move Thursday's launch review to 2:00 PM so we can include the final accessibility findings.

I'll share the updated checklist before noon.

Thanks,
Alex

Ready to insertEmail · On-device

One continuous workflow

Your cursor never loses the thread.

From first word to final insertion, Willy stays attached to the browser context where you started.

Put the cursor where you write

Stay in Gmail, LinkedIn, Notion, your CRM, or the form already open.

Start from the field

Use the hotkey, toolbar control, or in-field microphone. No app switching.

Speak naturally

Pause, restart, or change direction. Willy removes the verbal scaffolding.

Review finished text

The final result returns to the field in the format and language you chose.

The complete extension

More than a microphone button.

A browser-specific writing system with explicit controls for capture, terminology, transformation, insertion, and recall.

Capture

Starts where your cursor is

Use the keyboard shortcut, toolbar control, or in-field microphone. The extension remembers the field and site where you began, even if you switch tabs while speaking.

Transcribe

A real local model, not a privacy label

Whisper large-v3-turbo runs in the extension with WebGPU and a WASM fallback. The free and Lifetime paths do not upload audio for transcription.

Shape

12 formats, 54 translation targets

Clean text, email, notes, tasks, summaries, social and career workflows — plus direct translation into 54 languages. Set one globally or pin a different choice to a site or named field.

Insert

One final write, never a rewrite race

Willy inserts only the finished result. Gmail and rich editors get dedicated strategies; when a site refuses synthetic input, the text remains available to copy.

Recall

History stays in your browser

The side panel keeps up to 500 recent dictations in local IndexedDB. Copy, download, delete, or re-style from the original transcript.

Control

Choose who styles your words

Lifetime supports your own Claude, OpenAI, or Gemini key. Managed uses Willy’s account. Without either, local transcription still works.

A visible data path

“On-device” is an architecture decision.

In Free and Lifetime mode the whole pipeline runs in your browser: microphone → voice activity detection → Whisper large-v3-turbo via WebGPU. No audio, no transcript, and no account leave your machine. There is no server to trust because there is no server in the path.

On our own 20-clip benchmark (2026-07-12) that in-browser model transcribed at 0.0% word error rate, against 1.3% for OpenAI's whisper-1 on the same clips. That is our measurement on our clips, not a universal score — but it is why we can put accuracy and privacy in the same sentence.

Free / Lifetime

Your browser

Audio stays on device. WebGPU where supported; a slower WASM fallback elsewhere. First run downloads the model.

No network path

Managed

Willy → OpenAI

Audio passes through our proxy to the transcription provider. Willy keeps no copy; the provider's retention policy applies.

Cloud path disclosed

Lifetime BYOK

Your chosen provider

Provider keys stay in chrome.storage.local, unencrypted at rest. Use a restricted key and protected OS account.

You choose the provider

An honest comparison

Focused by design. Clear about the tradeoffs.

Willy is not the broadest dictation product. It is built for browser writers who want local transcription, visible data paths, and a low-cost purchase they can own.

Published pricing and product differences for Willy Dictation and three alternatives
Product	Published price	Local processing	Product shape	Best fit
Willy Dictation	$0 · $49 once · $9/mo	Yes — on-device STT is the default	Chrome / Edge extension	Private browser dictation with a low-cost ownership option
VoxWrite	Free · €39.99 once · €7.99/mo	Lifetime requires external AI API keys and usage costs	Chrome / Edge extension	A mature extension with custom rules and translation
Wispr Flow	Free · $15/mo ($12/mo annual)	Cloud product with a model-training opt-out	Mac, Windows, iOS, Android	Cross-device dictation and teams
Superwhisper	Free · $8.49/mo · $249.99 once	Local voice models are included in Pro	Mac, Windows, iPhone, iPad	Desktop power users who want modes, vocabulary, and meetings

Competitor facts checked against each company’s own published pages on August 11, 2026. Prices exclude taxes and can change; follow the linked product names for the current offer. The comparison scrolls horizontally on compact tablet layouts and becomes a stacked list on phones.

Choose Willy when…You mainly dictate in the browser, want audio-local transcription, and prefer $49 once or a $9 managed plan.

Choose Wispr Flow when…You need one polished service across desktop and mobile, or enterprise controls matter more than local processing.

Choose Superwhisper when…You want a broad desktop power-user suite, meeting workflows, and local custom vocabulary—and the higher lifetime price is acceptable.

Simple, explicit pricing

Choose who runs the intelligence.

Start on your own hardware. Buy once and bring an AI key, or use Willy's managed service.

Free

$0forever

On-device dictation, no account, no upload.

Whisper large-v3-turbo in your browser (WebGPU, WASM fallback)
Audio never leaves your machine
Raw punctuated transcript — add your own LLM key for styling
First run downloads the model (tens–hundreds of MB)

Store release pendingChrome and Edge listings are in review.

Lifetime (BYOK)

$49one-time

Own it. Bring your own AI key.

Everything in Free, plus the full styling pass
Your own Claude / OpenAI / Gemini key — you pay the provider directly
12 output formats plus direct translation into 54 languages
Offline Ed25519 licence — activate once, no server check afterwards
Keys are stored in your browser profile (chrome.storage.local), unencrypted at rest

Checkout opens soon.

Managed for you

Managed

$9per month · 7-day free trial

No keys, no setup. We run the models.

Cloud transcription and styling through our proxy — no API key to manage
Our keys, our bill; audio goes through Willy to OpenAI, and Willy stores no copy
Fair-use cap: 2,000 utterances or 500 audio minutes a month
Cancel any time from the billing portal
On-device mode stays available whenever you want it

Checkout opens soon.

Prices in USD. Buying requires a Willy account so the licence is issued to you — the extension itself runs without one. The managed tier bills monthly after a 7-day trial and can be cancelled from the billing portal at any time; the Lifetime licence is a one-time purchase and keeps working offline. Buying accepts the terms of service and the privacy policy.

Chrome + Edge

Put your voice where the cursor is.

The extension is in review for the Chrome Web Store and Edge Add-ons. A purchase made now links the licence to your account and activates when you install.

Open your account

Technical requirements

Know before you install.

Specific requirements and limitations, stated before purchase.

What it needs

Chrome or Edge 116+; Lifetime licence activation needs browser Ed25519 support (Chrome 137+).
A microphone and a first-run local model download. WebGPU is fastest; WASM is the fallback.
Access to text fields on the sites where you use it. Willy stays inert until you start dictation.

What it does not promise

Google Docs and some protected editors may require copy and paste when they reject browser-generated input.
Managed audio leaves the device, passes through Willy, and is forwarded to OpenAI. OpenAI’s retention policy applies.
BYOK provider keys live unencrypted in your browser profile. Use a restricted key and a protected OS account.