Text to Speech
Read any text aloud using your browser's built-in voices.
This tool runs entirely in your browser. Your data is never uploaded, never stored, and never leaves your device.
Reads text aloud through the voices already installed on your computer or phone — a text to speech reader, or TTS, using your system's own speech synthesis rather than a server — with adjustable rate and pitch and full pause, resume and stop control.
How to use it
- 1Type or paste into Text to speak.
- 2Choose a Voice — the dropdown lists each one's name and language, with your system default marked.
- 3Drag the Rate (0.5–2) and Pitch (0–2) sliders, then press Speak; Pause, Resume and Stop become available while it runs.
Example
- Input
- The quick brown fox jumps over the lazy dog — rate 1.4, pitch 1.0
- Output
- the sentence spoken in the selected voice, with a Speaking... label beside the buttons until it finishes
Voices are supplied by the operating system, so the list on a Mac, a Windows machine and an Android phone will not match, and a page loaded before the browser has finished registering them briefly shows "No voices loaded yet" until it refills the dropdown.
What happens to your data
The page hands your text to the browser's own speechSynthesis engine as an utterance object and makes no request of its own. Be aware of what that engine does next, though: some system voices are synthesised on the device and some are network-backed by the browser or OS vendor, so the voice you select — not this page — determines whether the words leave your machine. Leaving the page cancels any speech in progress, and the text is never stored.
Last updated August 2026
You have something you would rather hear than read — an essay you have looked at so often your eyes skate over the typos, a message to check before it goes, a chapter to get through while the dinner cooks. Listening catches what reading misses, because a clumsy sentence sounds clumsy in a way it does not always look.
Before you paste anything, work out which of two kinds of text to speech you want. One is playback: a voice reads the words now, through your speakers, and when it stops nothing is left behind. The other is production: you want an audio file, an MP3 for a video or a slide deck, and that means a service that renders speech to a file, usually with an account and a bill attached. This is the first kind — no export, no MP3, no recording button.
The other thing to settle is the voice, because it decides more than the tool does. Every sound you hear comes from a voice your operating system already holds, so the gap between a 1998 satnav and something close to a person is a question of what is installed on your machine. Most systems ship with a plain default and keep the better ones behind a download in settings.
The common mistake is pasting in a whole chapter and walking away. Your text goes to the engine as one utterance rather than being cut into sentences, and long passages are where browser speech is least dependable. Work a few paragraphs at a time.
How it works
Toolvore hands your text to the Web Speech API the browser already carries, as one utterance object with the chosen voice, rate and pitch attached, and leaves the browser and the operating system to do the synthesis. No audio is generated by the page, which is why the same words sound different on a laptop and on a phone. Pause, Resume and Stop are thin covers over the speech engine's own pause, resume and cancel, and pressing Speak cancels anything already running first, so two readings never overlap. The weaknesses follow from leaning on the platform: the whole box goes over as a single utterance rather than split at sentence boundaries, and Chrome in particular has a long-standing habit of falling silent partway through a long one. When synthesis fails, the failure handler only returns the buttons to their resting state — nothing on screen says what went wrong. There is no volume control, no language setting separate from the voice and no markup for pronunciation.
Common use cases
- Proofreading an essay by ear before you hand it in
- Hearing a long message read back to you before you send it
- Listening to notes or an article while you cook or fold washing
- Checking how a name or an opening line sounds spoken
- Reading a passage aloud for someone with low vision or dyslexia
- Comparing the voices on your machine before recording a script elsewhere
Frequently asked questions
How do I save text to speech as an MP3?＋
Not from here, and the reason is structural. Browser speech synthesis sends audio straight to your sound output; there is no point in that path where a page can catch the samples and package them, which is why no download button exists here. Three routes get you a file. Record what your machine is playing, using Audacity on Windows or a loopback device on macOS, which works but picks up every other noise your system makes. Use a cloud text to speech service, which renders to a file by design and charges by the character. Or use a desktop narration application.
Why do the voices sound so robotic, and how do I get better ones?＋
The list is whatever your operating system has installed, and the flattest voices are usually the defaults sitting at the top of it. On macOS and iOS, System Settings, Accessibility, Spoken Content, System Voice, Manage Voices opens a far longer list, and the entries marked Enhanced or Premium are large downloads that sound markedly better than the compact ones. On Windows, look under Settings, Time and Language, Speech, though Windows exposes only some installed voices to browsers. On Android the engine sits under Settings, Accessibility, Text-to-speech output. Reload the page afterwards, since the voice list is read as the page loads.
Why does the speech stop partway through a long passage?＋
Long passages are where browser speech engines are least dependable. The text is handed over as a single utterance rather than cut into sentences, and Chrome has a long-standing bug in which one long utterance stops with no error raised. Here that shows as the buttons returning to their resting state and nothing on screen explaining it, because the failure handler only resets the status. The fix is to work in smaller pieces, a few paragraphs at a time — which also makes it easier to hear one paragraph twice without hunting for your place.
Why does it mispronounce names, abbreviations and numbers?＋
Speech engines guess from spelling, and they guess badly on names, initialisms and anything numeric. Browser speech synthesis takes no pronunciation dictionary and no markup, so the only lever is the text itself. Respell the awkward word the way it sounds — Siobhan as Shivorn — in a copy you are listening to, never in the one you will send. Put full stops between the letters of an initialism to force them apart. Write a number out in words when the reading matters, since a bare 1996 may come out as a year or as four digits.
Is text to speech private, and where does my text actually go?＋
The page itself sends nothing and stores nothing — your text lives in the tab and goes when it does. The open question is the voice, since some are synthesised on your device and others are rendered over the network by the browser or OS vendor. There is a test that settles it: turn off Wi-Fi and press Speak. If it still talks, that voice is local. On macOS the downloaded Enhanced and Premium voices work offline, and on Android the Google engine can be pointed at installed data only. Run that test once with the voice you use for confidential text.
What is the difference between text to speech and a screen reader?＋
A text to speech reader speaks text you hand it. A screen reader — VoiceOver, NVDA, JAWS, TalkBack — is assistive software that speaks the whole interface: it announces buttons, headings, form labels and states, and it drives navigation by keyboard or gesture so someone can use a computer without seeing it. They overlap in that both end in a synthetic voice, often the same voice, but they are not substitutes. If you want a passage read out, a reader like this is enough; if someone cannot see the screen, they need the screen reader.
Can I use a computer voice in a video or a podcast?＋
Two separate problems, and the practical one comes first: there is no export here, so any published use means recording the playback or rendering the audio elsewhere. The other is licensing. Voices installed with an operating system are generally licensed for you to listen to rather than to publish, and Apple's and Microsoft's terms have historically drawn a line at redistributing the output. Some third-party voices are sold with a commercial licence attached precisely because that line exists. Check the terms for the specific voice rather than assuming that a system voice is free to broadcast.
How do I get a PDF or a web page read aloud?＋
Copy and paste is the honest answer for this tool: it takes typed or pasted text and offers no file picker and no way to point it at an address. For a PDF, most viewers already have something built in — Preview on macOS reads a selection aloud, Acrobat has Read Out Loud under View, and Edge will read a PDF with Read Aloud. For a web page, Edge and Safari both have a reader mode that speaks the article. A scanned PDF is where all of them fall down: it holds only a picture of text, so it needs OCR first.
