Turn any script into a natural-sounding voiceover — for videos, presentations, podcasts, or proofreading your own writing by ear. The AI voice runs entirely on your device and exports MP3 or WAV.
A one-time ~90 MB download, cached by your browser — afterwards it loads instantly and works offline.
Voices speak English. Numbers, dates, and common abbreviations are read naturally.
Every other text-to-speech site sends your script to a server, meters your characters, and watermarks or paywalls the download. This one doesn't, because it can't: the AI voice — the open-source Kokoro model — downloads once into your browser's cache and synthesizes speech right on your device. Type, listen, download. No account, no character limit, no upload.
That makes it safe for scripts you'd never paste into a cloud service: internal announcements, unreleased product copy, personal messages, or a chapter of the novel you're not ready to share.
Paste the text, pick one of ten voices, set the speed, and generate: play it instantly or download MP3 or WAV. No character counter ticking toward a paywall.
Because the voices are generated on your device rather than by a metered API. Paste a paragraph or a whole chapter; your machine does the work either way.
Yes: download the MP3 and lay it onto footage (our Audio Overlay tool does exactly that), which turns a silent screen recording into a narrated walkthrough without recording your own voice.
They are modern neural voices, well past the robot era: natural enough for narration, explainers, and listening to your own drafts read back, which remains the fastest way to catch clunky sentences.
No. The voices run locally after a one-time download, so scripts and drafts never leave your machine.