Text to Speech
Hear any text read aloud instantly, or turn it into a natural-sounding MP3 with an AI voice that runs on your device.
How to Use Text to Speech in 3 Steps
Type or paste text
Enter anything from a sentence to a few pages in the left pane.
Listen instantly
Press Speak to hear it with a voice built into your device. Pick a voice and speed in Settings.
Create an audio file
Click Create audio file for a natural AI voice saved as MP3 or WAV. The model downloads once, then works offline.
Text to Speech — Frequently Asked Questions
Is my text sent to a server?
No. Speak uses your operating system's voices, and the AI voice model runs inside your browser. The model files are downloaded from Hugging Face the first time; your text never leaves the device.
Why can't I download the Speak voice?
Browsers play device voices directly to the speakers and give web pages no way to record them. That is why audio files are made with the on-device AI voice instead.
How big is the AI voice download?
About 90 MB (Kokoro-82M, quantised), downloaded once and cached by your browser. A WebGPU-capable browser generates faster; otherwise it runs on the CPU at roughly real time.
Which languages are supported?
Speak supports every language your device has voices for. The AI voices are English (US and UK accents); other languages will be read with English pronunciation.
Is there a length limit?
Long text is split into sentences and generated piece by piece, so several pages work — it just takes longer. Very long jobs are easier in chunks of a few thousand characters.