yappy

local · 31 languages · open source (mit)

Any text, read aloud.

Share an article, a PDF, an EPUB, a Word file or a video, and Yappy turns it into a natural voice you can listen to while you cook, walk, or rest your eyes. The model runs entirely on your device: nothing ever leaves it.

generated on an ordinary laptop, no cloud, in seconds

In 1985, a letter arrived forty years late. Henry VIII never read it, but the 21st century did.

The highlighted parts are verbalized by the scriptwriter: "1985" becomes "nineteen eighty-five" and "Henry VIII" becomes "Henry the Eighth", with each language's own rules.

why it exists

No cloud, no accounts, no per-character fee.

Nothing leaves your device

Speech is synthesized on your computer or iPhone by an open ~66-million-parameter model. It works offline too: forever.

31 real languages

Ten voices, each of them polyglot. Documents with quotes in another language switch on their own, paragraph by paragraph.

Share anything

An article from Safari, a YouTube video, a PDF from your email, a voice note. Everything lands in the queue, gets prepared, and gets listened to.

the piece nobody else shows

The scriptwriter: from written to spoken.

Reading aloud is not spelling out. "1492" is said "fourteen ninety-two"; "Henry VIII" is "Henry the Eighth", but "Louis XIV" in French is "Louis quatorze"; "EL CID" is not a Roman numeral; the FBI is spelled out and NASA is not. Yappy turns every text into a reading script with each language's rules, and this box proves it: it runs the same Rust code as the app, compiled for your browser.

The script will appear here, with every transformation highlighted.

everything that comes in

Sharing is the door.

Articles

The page cleans itself: menus, banners and footers out. The prose stays, with its headings and its rhythm.

YouTube videos

The transcript comes along and reads like an article. No ads in between.

PDF and EPUB

With navigable chapters and export to a real .m4b audiobook, bookmarks included.

Word and ODT

Headings, lists and quotes keep their structure, and the scriptwriter gives them their breathing.

Voice notes

The local ear (Parakeet) transcribes them on-device and leaves them readable and listenable.

Whatever you're looking at

On desktop, one shortcut reads your selection, the browser page, or the active document. One second and it speaks.

free, mit, no sign-up

Downloads

First launch downloads the voice model (~380 MB) once. After that, Yappy works offline forever.

questions that arrive

Questions

Is it really free?

Yes. Yappy is open source under the MIT license. Being able to hear what you read is an accessibility matter, and accessibility tools should not ask for a credit card.

Which voice model does it use?

Supertonic 3, an open model by Supertone of about 66 million parameters that fits in 380 MB and synthesizes faster than real time on any laptop from the last few years.

Do my texts go to a server?

No. Synthesis, transcription and article extraction happen on your device. The only network Yappy touches is the initial model download and, if you share a URL, the visit to that URL from your own machine.

Which languages does it speak?

Arabic, Bulgarian, Croatian, Czech, Danish, Dutch, English, Estonian, Finnish, French, German, Greek, Hindi, Hungarian, Indonesian, Italian, Japanese, Korean, Latvian, Lithuanian, Polish, Portuguese, Romanian, Russian, Slovak, Slovenian, Spanish, Swedish, Turkish, Ukrainian and Vietnamese.