Why read a draft aloud
Your eye skips what your ear will not. Reading your own writing silently, you supply the punctuation you meant, not the punctuation you typed. A synthetic voice extends you no such courtesy — where it stumbles is usually where the sentence is wrong.
It is the fastest edit there is, and it catches a class of problem no spellchecker will: the sentence that is grammatical and still unreadable.
Where the voice comes from
This page does not ship a voice. It asks your browser to speak, using the voices your operating system and browser already provide — which is why the list differs between Windows, macOS, Android, iOS and Linux, and between Chrome, Safari and Firefox.
Voices come in two kinds, and the difference matters:
- Device voices run entirely on your machine. Nothing is transmitted. These are marked plainly in the list.
- Network voices are generated by the browser vendor’s servers. They usually sound better, and the text you are having read is sent to that provider to produce the audio. The list labels these as network voices so you can decide.
If the text is sensitive, pick a device voice.
Using the player
Paste the text, press play. While it reads, the box becomes a transcript and the word being spoken is highlighted, so you can follow with your eye and catch the places where what you hear and what you meant come apart. The bar shows how far through it is, with elapsed time on the left and an estimate of what is left on the right.
- Play and pause — the round button, or the space bar. Escape stops and returns to the beginning.
- Speed — 0.75× is close to careful speaking pace and catches more than 1× when proofreading. 1.5× and 2× are for getting through something you have already read once.
- Pitch — lower can be easier to listen to for a long stretch; it is remembered along with the speed and the voice.
- Stop — hands the text back so you can edit it. While it is reading, the text is locked, because a transcript that no longer matches the audio helps nobody.
Long text is split into sentences and queued rather than sent as one block. That is deliberate: Chrome silently stops a single long utterance after about fifteen seconds, which is why a page like this can seem to work on a paragraph and fail on an article.
Matching the voice to the text
An English voice reading French will mangle it. The language code sits next to each voice name — pick one that matches, and if your system has few, most operating systems let you add more under accessibility or language settings.
Common questions
Why is the voice list empty or very short?
Voices load asynchronously and some browsers populate the list a moment after the page. If it stays short, your operating system has few installed — most systems let you add more in accessibility or language settings.
Can I download an MP3?
No, and not because it was left out. The browser’s speech engine plays sound straight to your speakers — it never hands the page an audio stream, so there is nothing here to capture and save. The alternatives are both worse than the gap: a screen-recording prompt that on most systems records the wrong thing, or sending your text to a server to be turned into a file, which is exactly what this tool exists not to do.
If you need an audio file, your operating system may already do it — macOS can save spoken text to an audio file from the Services menu, and most systems have a screen recorder that captures system audio.
Why does the highlighting sometimes move a sentence at a time?
Most voices report where they are word by word, and the highlight follows exactly. A few — usually on Android — report nothing at all, so the tool falls back to marking the sentence being read rather than pretending to a precision it does not have.
Does it work offline?
Device voices do, once the page is loaded. Network voices need a connection by definition.
Can my phone do this to my own documents?
Yes — Wordpad Plus reads any document aloud on Android, highlighting the sentence it is on so you can follow and fix as it goes.
