Free Text to Speech
Text to Speech reads English text aloud with 24 American and British voices from the open-source Kokoro-82M model, and saves it as MP3 or WAV. Download the voice model once: 92 MB for the CPU or 326 MB for WebGPU. Numbers, dates and prices are spoken as words, and the current sentence is highlighted as it plays.
- Voices:
- 24: 16 American, 8 British
- Language:
- English only
- Output:
- MP3 or WAV download
- Price:
- Free
- Runs:
- AI model in your browser
- Uploads:
- None: files never leave your device
- Account:
- Not needed
Quick answers
Is the text sent to a server?
No. The voice model is downloaded once from Hugging Face and runs in your browser; your text never leaves your device.
Which voices are available?
24 English voices of the Kokoro-82M model: 16 American and 8 British, female and male. The grade next to each name reflects the quality and amount of its training data.
Processed on your device: your files are never uploaded. How it works
Network activity in this tab
What this page has requested since you opened the site in this tab. An upload is any request that sends data: POST, PUT, PATCH or a request with a body.
No uploads: nothing has been sent from this page.
No network requests yet.
To see every request, open your browser's developer tools (Network tab).
How to use the Text to Speech
- Type or paste English text (numbers, dates and prices are read out in words).
- Choose an American or British voice and the speed; click Download voice model the first time.
- Press Play: speech starts after the first sentence and the current sentence is highlighted. Click any sentence to start from it.
- Download the whole text as MP3 or WAV, or use Browser voices for an instant preview.
Credits & open source
Open-source AI model running in your browser. Thank you to the people who make these projects.
- Transformers.jsRuns Hugging Face AI models in the browser with ONNX Runtime. © Hugging Face, Inc. and contributors · Apache-2.0
- ONNX Runtime WebExecutes ONNX models with WebGPU or WebAssembly (self-hosted from /vendor/ort). © Microsoft Corporation · MIT
- CMU Pronouncing DictionaryEnglish word pronunciations (ARPAbet) used for text-to-speech phonemes; the dictionary itself is BSD-2-Clause. © Carnegie Mellon University (dictionary); package by Zeke Sikelianos · ISC
- lamejsMP3 encoder (port of LAME), loaded as a separate unmodified chunk. © LAME project; JavaScript port by Alex Zhukov; ESM fork by Breezystack · LGPL-3.0
Web platform APIs used: Web Workers, WebAssembly, Web Audio API, localStorage, File API.
- Kokoro-82M (ONNX export, revision 1939ad2) by hexgrad (model); onnx-community (ONNX conversion) (Apache-2.0) — neural voice model: phonemes to 24 kHz speech (8-bit weights on CPU, 32-bit on WebGPU)
- Kokoro-82M voice style vectors (af_heart, af_bella, bf_emma, bm_george and 20 more) by hexgrad (Apache-2.0) — American and British English voices (downloaded on demand, ~0.5 MB each)
- CMU Pronouncing Dictionary by Carnegie Mellon University (BSD-2-Clause) — English word pronunciations (ARPAbet) for the phoneme front-end
Full license texts: third-party notices · all credits
Frequently asked questions
Can I use the audio commercially?
The Kokoro model and its voices are released under the Apache-2.0 license, which allows commercial use. Check any rights in the text you read.
How do I fix the pronunciation of a word?
Write it as [word](/phonemes/) with Kokoro's phoneme symbols, for example [Kokoro](/kˈOkəɹO/). Turn on Show phonemes to see how each sentence was read.
Why is it slow on my computer?
Without WebGPU the model runs on one CPU core. Chrome or Edge with WebGPU generate speech many times faster than real time.