EnviousWispr
GitHubDownload
Self-Learning Dictionary

A personal dictionary
that builds itself from your edits.

Edit a misrecognized word in a dictation EnviousWispr just pasted, and it detects the correction entirely on-device. No manual list maintenance, no cloud sync required. It learns your unique vocabulary seamlessly in the background.

Download for MacFree for Mac · macOS 14+ · Apple Silicon

Correcting a name

On your Mac
Pasted a moment ago

Send the draft to SashaSaoirse tonight.

Added “Saoirse” to DictionaryUndo
Your next dictation

Saoirse signed off on the draft.

Checked locally each time before changing your text.
1 of 3
Illustrative example
Everyday editing

Just edit your text
the way you normally would.

Dictate something, let EnviousWispr paste it, and edit whatever is wrong. When EnviousWispr spots a correction, a small pill shows what it saved, such as Added “Saoirse” to Dictionary, or “Notion” updated for a word you already had, with an Undo button for four seconds. If you leave it alone, it stays. New words appear in Your Words, while existing words simply log the new mishearing.

Four seconds to undo

If it caught something you did not intend, click Undo on the pill. That removes a freshly added word or reverts an updated one back to where it was.

Marked in your dictionary

You can see what was picked up automatically at any time. Words added this way have a small sparkle icon in Your Words, and the Auto-learned filter lets you view just those.

Supported apps

It watches for corrections in Mail, your web browser, terminal windows, and web-based apps like WhatsApp. Password fields are never watched.

Context checking

Context-aware replacements,
verified locally.

Traditional custom dictionaries have a major flaw: map "twist" to "Tuist," and you can never use the word "twist" normally again.

For words it learned from your edits, with Enable Dictionary on, EnviousWispr resolves phonetic collisions by looking at context. It evaluates both words against the sentence to know when you are talking about Xcode build systems and when you are talking about opening a jar.

The build runs on Tuist now.

Changed
=

Give the lid a twist.

Left as you said it
Learned word: Tuist, heard as “twist”. Illustrative example.

Your corrections stay
on your computer.

Detecting corrections and checking context happen entirely on your Mac, regardless of your polish settings. Envious Labs never receives the text being watched, the word you fixed, or the mishearing. If you choose cloud polish, your dictation, with any learned words in it, goes directly to your chosen provider under your personal key.

Read how it works, which apps it watches and how to turn it off
A little more room for your ideas.

The dictation app that actually gets your names and jargon.

Free, private dictation for Mac. No account. No subscription.

Download for MacmacOS 14+ · Apple Silicon