Text to Speech

How to use Text to Speech

1. Paste or type your text.
2. Choose a voice from the ones your device provides.
3. Adjust rate and pitch, then play.
4. Available voices depend entirely on your operating system and browser.
5. The text is not transmitted — synthesis happens on your device.
6. There is no download: the API speaks rather than producing a file.

The voices come from your device, not from here

This uses the Web Speech API, which hands your text to whatever synthesiser the operating system provides. That means the available voices differ completely between a Mac, a Windows machine, an Android phone and an iPhone, and a voice you choose here will not exist for someone on another platform. It also means quality varies: recent macOS and Android voices are very good, older Windows ones noticeably robotic.

What it is genuinely useful for

Proofreading is the best use by a distance — hearing your own writing read back catches clumsy sentences, repeated words and missing ones that your eye skips over, because you wrote them and know what they should say. Beyond that: listening to an article while doing something else, checking how a name or a brand is pronounced, and accessibility support for reading difficulties, where hearing alongside reading measurably helps comprehension.

Rate, pitch and the limits of tuning

Most people prefer a rate slightly above the default once they are used to it, and regular listeners often run at 1.5× or more. Pitch adjustment is cruder and usually makes a voice sound worse rather than different. What you cannot fix with these controls is prosody: device voices place emphasis mechanically, so they mispronounce homographs, flatten questions and stumble on names. For anything published, record a person.

It is speech, not an audio file

The API speaks; it does not hand you a file. Saving the output means capturing it with recording software, and the licence terms on many system voices restrict redistributing their output. For a podcast, a video voiceover or anything commercial, use a service that licenses its voices for that purpose — and be aware that several now require disclosure when synthetic speech is presented as a person.