Text to speech

This text-to-speech reader uses the speech synthesis built into your browser — the Web Speech API — so it works without an account, a download or any audio being streamed from this site. You paste the text, choose from the voices your device already has, and the browser speaks it. Because the voices belong to your operating system and browser, the choice ranges from a handful on a basic Linux install to well over a hundred on a recent iPhone or Mac, covering dozens of languages. The list is grouped by language with your own first, and each voice shows its region, so an American and a British English voice are easy to tell apart.

Speed runs from half to double the normal rate, pitch from 0 to 2, and volume from silent to full; how much a voice actually responds to pitch varies between engines. Changes made while the text is being read take effect from the current word, since a sentence that has already been handed to the speech engine cannot be altered. Your voice and settings are saved in this browser, so the next visit starts where you left off.

While it reads, the text turns into a reading view that follows along: the current sentence is shaded, the current word is highlighted when the voice reports word positions, and the view scrolls to keep your place visible — useful for proofreading a draft by ear, following along in a language you are learning, or listening to a long article with your eyes resting. Pause remembers the last word spoken rather than relying on the browser’s own pause, which is unreliable on Android and with online voices, so Resume continues from the highlighted word on every platform.

A few limits come with the browser API. The audio cannot be saved as a file, because browsers never expose it to the page. Pronunciation of names, abbreviations and numbers is decided by the voice, not by this tool, and a voice only reads its own language well — pick a German voice for German text. Text is capped at 100,000 characters, roughly two hours of listening at normal speed.