Speak your own language, paste English prompts

A lot of developers think in one language and work in another. You can read English documentation all day, but when you're explaining a tricky bug, the explanation comes out faster and clearer in Hindi, Urdu, Spanish, Arabic or whatever you grew up speaking.

Most AI coding tools work best with English prompts. So you end up translating in your head, which is slow and loses detail.

There's a simpler way: speak in your language, and paste English.

How it works in Whisp

1. Pick your spoken language in Whisp. ⌥L opens a quick picker with the languages installed on your Mac.
2. Press ⌥ Space and talk normally.
3. Instead of Return, press ⌥Return to finish.

Whisp transcribes what you said in your language, translates it to English on your Mac using Apple's translation framework, and copies the English for ⌘V. Both versions are saved in your history, so you can always go back to what you actually said.

The first time, Whisp can prepare the language pair from Settings so macOS downloads what it needs. After that, the translation runs on your device.

Code terms survive the translation

This is the part that usually breaks. Translation engines are built for sentences, and they happily "translate" a file name or mangle a command.

Before translating, Whisp sets aside the things that shouldn't change: file names like package.json and .env.local, URLs, anything in backticks, camelCase and snake_case identifiers, common commands like git status or npm run dev, and the custom vocabulary you've added. It puts them back after translating, and if any of them went missing, it keeps your original transcript instead of handing you a broken one.

When this is useful

• Explaining a bug in detail, where nuance matters.
• Long architecture notes you'd struggle to phrase in English quickly.
• Pairing with a teammate in your shared language, then pasting a clean English summary into an issue.

It's less useful for very short prompts. If you can type it in English in five seconds, just type it.

Things to know

• Recognition quality depends on the on-device speech model macOS provides for your language. Some languages are stronger than others.
• One recording uses one spoken language. If you mix languages a lot mid-sentence, results vary.
• English output is cleaned up after translation, so punctuation and file names still come out right.

Try it

Whisp has a seven-day free trial on the Mac App Store, no card needed, then $9.99 once. Speech and translation stay on your Mac.