AI text to speech uses the Kokoro neural network to generate natural, expressive English speech from text. Choose from 30+ voices - American and British English - each with its own personality and tone.
Everything runs locally in your browser via ONNX Runtime - your text never leaves your device. The first run downloads a ~170MB model, cached for instant reuse. Export the result as a WAV file. Prefer zero download? Pick the native system voice for 30+ languages.
Yes - completely free with no limits. The AI runs on your own device.
No. The model downloads to your browser and all generation happens locally.
30+ AI voices: American English and British English (male and female).
A WAV audio file of the generated speech.
English with the AI voices; 30+ languages (including Chinese) via the native system voice option.