What does a text to speech reader do?
It takes written text and speaks it aloud using the speech voices your operating system already has installed. The browser exposes them through the Web Speech API, so the reading happens through the same engines a screen reader uses, and there is nothing to install and no account to create.
The controls are the ones that matter for listening rather than for synthesis. Speed, pitch and volume are adjustable while the reading is stopped, the voice list can be filtered down to English or left showing everything installed, and the current word or sentence is highlighted as it is spoken so the eye can follow the ear.
The one thing that cannot be done from a web page is worth stating plainly: this tool cannot give you a downloadable audio file. The API plays sound; it does not hand back a buffer you could save as an MP3. Anything advertised as downloading speech from a browser is either recording the system output on your own machine or sending your text to a server — this page does neither.
How to read text aloud
- Paste or type the text you want read into the box, or press Load an example to hear how the controls work.
- Choose a voice. Everything installed on the device is listed, with English voices grouped first, and each one notes whether it is local or synthesised over the network.
- Set the speed, pitch and volume. A speed between 0.9× and 1.25× sounds most natural for long passages; slower helps when the text is unfamiliar.
- Press Read aloud. The current word or sentence is highlighted in the panel beneath the controls as it is spoken.
- Use Pause, Resume and Stop while it reads, or Continue from here to pick up again from wherever the highlight had reached.
- Edit the text and the reading restarts from the top — nothing is read while you are typing.
Text to Speech limits and rules
| Item | Limit |
|---|---|
| Voices | Whatever the operating system ships, from none to several dozen |
| Local or network | Stated per voice; a network voice sends the text to the browser vendor to be synthesised |
| Controls | Speed from 0.5× to 2×, pitch from 0 to 2, volume from 0 to 100 per cent |
| Highlighting | Word or sentence boundaries as the engine reports them, which varies by voice |
| Audio file | Not producible — the API plays sound and returns no buffer to save |
| Browser support | Chrome, Edge and Safari on desktop; a browser with the API removed will say so rather than fail silently |
| Length | Very long text is handled by the engine, but some voices stop after a few thousand characters |
| Languages | Every language a voice is installed for, which follows the operating system |
When to use it
- Listening to a document or an article while doing something else.
- Checking how a paragraph sounds when it has to be read aloud in public.
- Hearing a long block of text read at speed to catch a sentence that does not parse.
- Confirming that accented characters and proper nouns are correct by listening to them.
- Reading text on screen when reading it visually is not comfortable or not possible.
Is it private?
The text is handed to your device's own speech engine and is not uploaded by this page. One caution that is stated rather than hidden: a voice marked as a network voice is synthesised on the browser vendor's servers, so with one of those selected the text you are reading is sent away — pick a local voice if that matters.
Frequently asked questions
Because the Web Speech synthesis API has no such output. It is designed to speak through the device's audio, so it takes text and produces sound, with no interface that returns a file or a buffer. A browser has nothing to hand you. Tools that offer a download either capture the system audio on your own machine, which is a desktop recorder rather than a web page, or run a server-side synthesiser and send your text to it. Both are legitimate, and neither is what this page does.
Because the voices belong to the operating system, not to the browser. Windows and macOS both ship a handful by default and let you add more; Chrome and Safari simply read that list. A Linux desktop often has none at all until a speech engine such as eSpeak or a Festival voice is installed. So the same page can offer thirty voices on one machine and none on another, which is why the count is shown — it is the fastest way to see whether the device has anything to read with.
It means the text is not synthesised on your computer. Chrome and Edge offer some voices that are generated by their vendor's cloud service, which is why they sound noticeably better and also why they need a connection. Selecting one sends the passage you are reading to a third party. That is a real difference from every other tool on this site, so it is labelled in the voice list rather than assumed. Choose a voice not marked network and the reading stays on the device.
Because engines report boundaries differently. The API allows a synthesiser to emit an event at each word boundary or only at each sentence, and the voice, not the page, decides which. Most modern voices report words; older and smaller ones report sentences or nothing at all. When nothing is reported, the reading is unaffected — the highlight is a courtesy, not the mechanism — and Continue from here falls back to restarting from the top of the passage.
Paste the text in and yes, provided you can select it, which is the catch. Text inside a scanned PDF is an image and cannot be selected at all; a page that blocks selection or serves text through a canvas has the same problem. For a scanned document, run it through an OCR tool first and paste the result here. For a web page that will not let you copy, most browsers have a reader mode that makes the text selectable.
There is no limit in the page itself, but individual voices are less reliable on long passages. Some engines stop partway through a few thousand characters, and a few cut off at the first paragraph break. If a document is long, reading it in sections is more dependable than pasting the whole thing, and the Continue from here button makes that practical: read the first part, then start again from wherever the highlight stopped.
Related tools
Written and maintained by Md. Nur Hossain Repon.