Text to Audio
Paste any amount of text and have it read aloud by the voices already installed on your device. What makes this one finish the job is the queue: the text is split into sentences, each kept to 240 characters or fewer unless a single word is longer than that, and spoken one at a time, so a long document reads to the end instead of stopping after the first paragraph the way a single long utterance does. The sentence being spoken is highlighted as it goes, reading can start at the top, at your cursor or where you last stopped, and everything runs in the browser.
0:00 / 0:00Stopped
Reading starts at the first sentence. The clock is an estimate at 150 words a minute scaled by the speed setting, because each voice sets its own pace and the speed control multiplies whatever that pace is.
Why this reads a whole book and other pages stop after a paragraph. Browsers cut a single utterance off after a few hundred characters, so handing the speech engine an entire document loses most of it. This splits the text into sentences and queues them one at a time, which is why a long document reads to the end and the highlight always shows where you are. The voice list is whatever your system and your browser provide, which is why quality differs between machines.
Where your text goes. This page has no server and uploads nothing. There is one honest exception, and the voice list above marks it: a voice that runs over the network is synthesised by the browser vendor, so the browser sends it the sentence it is about to speak. Choose a voice marked as running on this device and the text never leaves the machine at all.
Accuracy. Speech comes from the voices installed on your device, so quality and languages differ between machines. Text is split into sentences and queued so long documents read to the end. The browser does not hand the audio back, so there is no download; recording your speaker output is the workaround.
Common questions
- Why do other text to speech pages stop reading after a paragraph?
- Because they hand the whole document to the browser as one utterance, and browsers cut a long utterance off after a few hundred characters. There is no error, the voice simply stops. This tool splits the text at sentence ends into pieces of at most 240 characters and queues them, starting the next piece the moment the last one finishes, so the length of the document is not a limit. Sentences longer than that are broken at a comma, semicolon or colon, and failing that at a space, never inside a word: one word longer than the limit, a long URL say, is left whole rather than mispronounced in two halves.
- Why does the same text sound different on my phone and my laptop?
- Because the voices are not part of this page. They come from your operating system and your browser, so a Mac, a Windows PC and an Android phone each offer a different set, and the voices a platform marks natural, premium or enhanced are its newer and larger ones, which is where most of the difference in quality comes from. The voice list on the page shows everything your browser reports, with its language tag and whether it runs on the device or over the network.
- Can I download the audio as an MP3?
- Not from the built-in voices, and it is worth knowing why before you go looking for a setting. The browser speech API plays through the speakers and never hands the page the audio samples, so there is nothing here to save to a file. The workaround is to record your own speaker output with your operating system's recorder while the page reads, or to use a speech program that runs on your computer and writes files directly.
- Does my text get sent anywhere?
- This page has no server and uploads nothing: the splitting, the highlighting and the controls all run in your browser. One honest exception is worth naming. Some voices are network voices, synthesised by the browser vendor rather than on your machine, and choosing one means the browser sends it the sentence it is about to speak. The voice list marks which is which, so picking a voice that runs on this device keeps the text on the machine.
- How do I make it read another language?
- Open the voice list and pick a voice whose language tag matches your text. The tool starts on a voice that matches the page language, preferring ones the platform marks natural, premium or enhanced and ones that run on the device. If no voice for your language is installed, add one: on Windows under Settings, Time and language, Speech; on macOS under System Settings, Accessibility, Spoken Content; on Android by installing a text to speech engine, then reloading the page.
- Is there a limit on how much I can paste?
- None is imposed. A book-length paste is split into sentences the same way a paragraph is, and your position is remembered along with the text, so you can stop, come back, and start again from the sentence you left. Start from Selection reads from wherever your cursor sits, which is the quickest way into the middle of a long document.
- How close is the time estimate?
- Treat it as a guide. The clock counts 150 words a minute and divides by the speed setting, which is roughly a measured reading pace, but every voice has its own natural tempo and the speed control multiplies whatever that tempo is. The progress bar is the firmer number: it tracks the characters actually spoken, moving word by word where the browser reports word boundaries and sentence by sentence where it does not.
Speech comes from the voices installed on your device, so quality and languages differ between machines. Text is split into sentences and queued so long documents read to the end. The browser does not hand the audio back, so there is no download; recording your speaker output is the workaround.