Text to speech
Paste the text, pick a voice, press play. The words highlight as they are spoken, and nothing you paste goes anywhere.
Guide
How to use it
- Paste up to 5,000 characters.
- Pick a voice, the list is your device’s own installed voices, so phones, Macs and Windows machines each offer different ones.
- Set speed (1.0 is natural, 1.25 to 1.5 suits proofreading) and pitch to taste.
- Play. The current word highlights as it is read, pause and resume freely.
Examples
What reading aloud is for
Hearing your own writing read back catches what eyes skim past: missing words, doubled words, sentences that never end. At 1.25× speed it is faster than re-reading, and stricter.
Long articles become listenable. Paste, play, close your eyes, the highlight keeps your place for when you look back.
Device voice packs cover dozens of languages, paste French text, pick a French voice, and hear native-ish pronunciation at half speed.
Method
How it works
This uses the Web Speech API: the browser hands your text to the operating system’s speech engine, which is why the voice list is your device’s and why quality varies, recent phones carry genuinely good voices, older desktops carry robotic ones. Voices marked "cloud" in the list are synthesised by the OS vendor’s servers rather than locally, pick unmarked ones if that matters.
The word highlight follows the engine’s boundary events. And the honest limitation, stated plainly: a static page cannot record the OS engine’s output to a file, downloading speech as MP3 genuinely requires server-side synthesis, which is a different kind of product.
FAQ
Frequently asked questions
Why does my phone have better voices than my laptop?
Voice packs ship with the OS, and mobile platforms have invested most. Adding voices in your system settings (Windows: Speech; macOS: Spoken Content) upgrades this page instantly.
Can I download the audio?
Not from this page, and be wary of "free" sites promising it, they are uploading your text to their servers to synthesise. Local speech and downloadable files are different architectures.
Is there a length limit?
5,000 characters per read here, a comfortable article section. Some engines go quiet on very long runs, chunking long documents reads more reliably.
Why did the voice change mid-sentence?
Some systems substitute voices for languages the chosen voice cannot speak. Matching the voice’s language to the text avoids it.
Does it work offline?
With local voices, yes, once the page is loaded. Cloud-marked voices need a connection.
Is my text private?
It never reaches this site. Local voices process it on-device, cloud voices route it to your OS vendor (Apple, Google, Microsoft) under their terms, the voice list marks which is which.
More tools