Text to speech — natural voices, privately in your browser

Paste any text, pick a voice, and generate natural-sounding speech entirely on your device. Play it back and download a WAV. Nothing is uploaded — the model runs in your browser, so it's free, unlimited, and private.

  1. Your text

    0 characters · about 0:00

    Formatting — add pauses & pronunciation

    Type these anywhere in your text. They shape the speech but aren't read aloud.

    Pause

    Insert a real silence.

    [pause] · [pause 1.5s] · [pause 800ms]

    Pronunciation

    Respell a tricky name or acronym.

    [Nguyen](win) · [GIF](jif) · [Siobhan](shivawn)

  2. Voice & speed

  3. Speech

How to turn text into speech

  1. Paste your text

    Type or paste the text you want spoken. It never leaves your browser.

  2. Pick a voice and speed

    Choose from 30+ English voices (American and British, female and male) and set the speaking speed.

  3. Generate & download

    The voice model runs on your device (first run downloads it, then it's cached). Play the audio back and download it as a WAV file.

Why generate speech here

Your text is never uploaded

Speech is generated 100% in your browser — no server, no account, no upload. Ideal for scripts, drafts and anything confidential.

Natural, expressive voices

Powered by Kokoro, a compact but high-quality voice model, with 30+ English voices to choose from and adjustable speed.

Free & unlimited

Because synthesis happens on your device, there's no per-character cost and no cap. Generate as much speech as you like.

Questions

Is my text uploaded to a server?
No. The voice model is downloaded to your browser and runs on your own device — your text never leaves it.
What do I get to download?
A standard WAV audio file (mono, 24 kHz) that you can play anywhere or drop into a video editor.
Which voices and languages are supported?
30+ English voices at launch — American and British, female and male — powered by the Kokoro-82M model. More languages are planned.
Is it really free?
Yes — free and unlimited. There's no account and no metering because the model runs on your device, not our servers.
Why is the first run slower?
The first run downloads the voice model once, then it's cached in your browser, so later generations start quickly.
Does it need a fast computer?
It uses your GPU (via WebGPU) when available and falls back to your CPU otherwise. The model is small, so it runs on phones too — just a little slower.