Text
Text to Speech (Read Aloud)
Paste text and hear it read aloud — no sign-up, no character limit. The voice picker marks which voices run on your device and which send text to a server.
Nothing you paste here is saved — not on this device, not anywhere.
This page reads out loud. It cannot save an audio file, and no browser lets a page record one.
Waiting for this device's voices.
Reading the voices installed on this device.
These are the voices your device and browser already have. This page adds none of its own and buys none from a speech service.
Paste something you have written, pick a voice, and press Play. Sentences appear one at a time in a reading pane and the one currently being read is marked as it goes — with an underlined word too, on browsers that report timing accurately enough to trust. Tap any sentence to jump straight to it, whether the tool is mid-read, paused, or finished; that jump is what turns a one-way playback control into something worth coming back to.
Read text aloud
The controls are Play, Pause or Resume, and Stop, plus a Voice picker and two sliders for speed and pitch. Press Play and the text you pasted is split into sentence-sized chunks — long sentences are broken again at commas or other clause punctuation, never mid-word — and read back one chunk at a time, each one lasting roughly nine seconds of speech at most so a long paragraph never sits inside a single unbroken utterance. The chunk currently playing is highlighted in the reading pane; press any other row and reading jumps there immediately, cancelling whatever was queued. Rate runs from 0.5× to 2× and pitch from 0.5 to 1.5, both starting at the browser's own default, and changing either one mid-read restarts the current sentence at the new setting rather than waiting for a natural break.
This is also the accessibility use of the page, plainly stated: reading text aloud through the browser's own voices is exactly what a screen reader does, minus the rest of a screen reader's job of narrating buttons and menus. It is useful on a device where you don't have a screen reader turned on, don't want to change your system-wide settings just to hear one document, or are helping someone who finds listening easier than reading a wall of text on a small screen.
Proofread by listening
Reading your own writing out loud, or having something else do it, catches mistakes your eyes slide past — a repeated word, a dropped word, a sentence that runs on past the point where a reader's attention does. That is the actual job this tool is best at, and the sentence-by-sentence chunking above is what makes it work as a proofreading tool rather than just a playback button: each row in the reading pane is a distinct, independently replayable unit. Hear a sentence that sounds wrong, tap it again to confirm you heard it right, then press "Edit text" to swap the reading pane back for an editable box with your text intact. Fix the sentence, paste the corrected draft back in if you rewrote elsewhere, and press Play again. There is no separate "proofreading mode" to turn on — it's the same reading pane, used slowly and with the pause button close at hand.
Pronunciation practice
The same tap-to-replay row works for a single word or phrase, which is what makes this useful for practising how something is actually said. Paste the word, name, or sentence you want to hear, choose a voice whose language matches it — the voice list shows each option as its name and its language together, so you can tell "Daniel — English (United Kingdom)" apart from a voice in the language you're learning — and tap that one row as many times as you need. Slowing the rate toward 0.5× stretches the word out without distorting the vowels the way physically slowing down a recording would, which makes it easier to isolate a sound you're trying to copy. The voice list only ever shows what your device and browser already have installed, so if the language you're practising isn't offered, installing a voice for it through your operating system's language or speech settings is what adds it here — this page cannot add a voice that isn't already on your machine.
Which voices keep your text on your device
This is the one thing about the tool worth reading before you paste anything you'd rather not send elsewhere. The voice picker is grouped into three sections, and the group a voice sits in is not a formatting choice — it is a direct statement about where your pasted text goes:
- On this device — the voice runs locally. Your text is converted to speech inside your own browser and never leaves it.
- Over the internet — the voice runs on its maker's servers. Chrome's "Google …" voices and Edge's "Microsoft … Natural" voices are the common examples: choosing one sends your pasted text to that company to be turned into audio and streamed back, which is also why these tend to sound more natural than a local voice.
- Not stated by this browser — the browser does not report which of the above is true for this voice, so this page cannot promise either. Treat a voice in this group the way you'd treat an over-the-internet voice if privacy matters to you.
The tool defaults to an on-device voice where one exists, precisely so the safe choice is the one you get without opening the picker at all. A line under the picker names which group the currently selected voice is in every time you change it. That grouping is read from a flag the browser itself reports and is not perfectly reliable across every engine, which is exactly why the third group exists rather than being folded into one of the other two.
Why there's no download button
There is no way to save what this tool reads as an audio file, and there will not be one added later — it is a limit of what a browser lets a web page do, not something left unfinished. The sound goes from the speech engine straight to your speakers along a path a page cannot read from, record, or write anywhere. The only real workaround, recording your screen's system audio while the page plays, requires a permission dialog and does not work in Safari, in Firefox, on macOS, or on any mobile browser at all, so it would work for a minority of visitors and silently fail for everyone else — worse than not offering it. This tool reads text aloud in your browser; producing a file to keep is a different job, done by recording software on your own computer or by a paid speech service built for that purpose.
Questions
- Why does text to speech sound robotic?
- Because the voice you picked is most likely one that runs entirely on your device, and on-device voices trade naturalness for not sending anything anywhere. The smoother, more natural-sounding voices — Chrome's "Google …" voices and Edge's "Microsoft … Natural" voices — are synthesized on that company's own servers, which means your pasted text is sent there to be turned into audio. This tool defaults to an on-device voice for exactly that reason, and marks every voice with where it runs, so if you want the more natural sound and are comfortable with the text leaving your browser, open the voice picker and choose one under "Over the internet." A lower rate and a pitch close to 1.0 can also make an on-device voice easier to listen to, since pushing either slider to its extreme tends to make speech harder to follow, not easier.
- Why can't I download the audio as an MP3 or WAV file?
- Because there is nowhere in a browser for a page to catch it. The voice speaks directly from the speech engine to your speakers, and that path does not run through anything a web page can read, record, or write to a file — there is no audio graph a page can tap. The usual workaround, recording your screen with system audio while the page reads, needs a screen-share permission dialog and a checkbox most people miss, and it does not work at all in Safari, in Firefox, on macOS, or on any mobile browser, so shipping it here would work for a minority and fail silently for everyone else. If you genuinely need an audio file, the honest options are screen-recording software on your own computer, or a paid text-to-speech service built to produce one — this tool reads text aloud and stops there.
- Why are there no voices in my browser?
- Because this tool only ever offers the voices your operating system has already installed — it does not ship, download, or buy any of its own. If the box under Voice says your device has no speech voices installed, the browser can technically speak but the OS has not given it anything to speak with, which is most common on Linux and on stripped-down Android builds. The fix happens outside this page, in your operating system's language, speech, or accessibility settings — install a system voice there and this tool picks it up automatically the next time it checks, with no reload needed. If voices normally show and none appear here, wait a couple of seconds first: some browsers report an empty list on the very first check and fill it in shortly after.
- Is it safe to paste something private into a text-to-speech tool?
- It depends entirely on which voice is selected, and this is the one thing worth checking before you paste anything sensitive. Nothing you type is ever saved by this page — not to this device, not to an account, not anywhere — so closing the tab clears it completely. But the audio itself is a separate question: a voice marked "On this device" speaks locally and your text never leaves the browser, while a voice marked "Over the internet" sends your pasted text to that voice's maker to be converted into audio before it plays. A third group, "Not stated by this browser," covers voices where the browser will not say which is true, so this page cannot promise either way for them. For a cover letter, an unpublished draft, or anything about another person, pick a voice under On this device.
- Why did the highlighting stop working partway through reading?
- Most likely because your browser's speech engine gave up reporting exactly where it is. This tool underlines the current word using timing events the browser sends while it speaks, but not every engine sends them, and some send them with a position that does not actually match what is being read. Rather than show a wrong or drifting underline — worse than no underline at all — the tool switches word highlighting off for the rest of that reading once it sees three of those events land somewhere they shouldn't, and it does so without a message, because there's nothing you can do about it and a warning would just be noise. The sentence you're on keeps being marked the whole time regardless, since that position comes from the reading itself finishing a sentence rather than from those timing events, so you can always see roughly where playback is even with the word underline off.
- Why does it keep speaking for a few seconds after I press Pause?
- Because pausing mid-speech is not reliably supported by every browser's speech engine, and this is most common on Android. This tool tries to pause immediately, then checks a fraction of a second later whether the browser actually stopped. If it didn't, the tool stops queuing anything further and lets the sentence already playing finish naturally rather than cutting it off awkwardly or leaving you unsure whether it's frozen — a note appears saying it's finishing the current sentence before pausing. Either way, pressing Resume picks up exactly where reading stopped, at the start of the next sentence in this case, with nothing repeated and nothing skipped.