VoiceRuler is a free text-to-speech tool that runs the open-source Kokoro-82M voice model (Apache-2.0 licence, by hexgrad, converted to ONNX by the onnx-community project) inside your browser through kokoro-js. Your text is never uploaded to us.
How it works. The first time you press Read aloud, your browser downloads the voice model (about 90 MB) and the voices from this site, plus a code library from jsDelivr. They are cached, so later visits start faster. Speech is then generated on your own device, part by part, so you can start listening before a long text is finished.
Honest limits. The voices are English only (American and British). Names, abbreviations, numbers with units and unusual words can be mispronounced; spelling a word the way it sounds usually fixes it. Generation speed depends on your device: laptops are much faster than phones.
Who is behind it. VoiceRuler is published by Ecomparadise, LLC and edited by Alex. We build small, single-purpose tools, test them in real browsers, and explain plainly what they can and cannot do.
Found a bug or a mistake? Please tell us and we will fix it.