Text to Speech

Under “Where it runs” you choose: our server, where MeloTTS speaks it in seconds with nothing to download (the text isn’t stored), or this device, where nothing is uploaded and Kokoro (Apache 2.0, about 90 MB, downloaded once) speaks it with 13 voices and a speed you choose, using pronunciation dictionaries from the misaki project (Apache 2.0) and rules written for this site.

English, with American or British voices. Numbers, money, times and years are read out properly; unusual names are sounded out by rules and can be a little off. Other languages would need their own pronunciation rules, which this tool doesn’t have yet.

Up to 5,000 characters at a time, about five minutes of speech (and 5,000 characters a day on our server). It’s read in sentences, so long texts take a little while; the page shows how far it has got.

Yes. Kokoro is released under the Apache 2.0 licence and MeloTTS under the MIT licence, and the audio you make is yours to use.

Browsers can speak text, but they can’t save it as a file, and some send the text to an online service to do it. This tool makes the speech as a file you can keep, on our server or on your device.

It is free, with no watermarks and nothing to sign up for. On this device there are no daily limits. Our server has a free daily allowance, and a free account makes it six times bigger.