Skip to content

Ultra-lightweight 15M-param English TTS — only a 24MB quantized download, the smallest and fastest-loading model in the lab. 8 voices phonemized via eSpeak NG WASM. WASM-only: the repo ships a single q8 ONNX file with no fp32/WebGPU variant.

WASM
Apache-2.0

Embed Kitten TTS Nano

Copy the code below to embed this TTS demo on your website.

<iframe src="https://ttslab.dev/embed/kitten-tts-nano" width="100%" height="400" frameborder="0" allow="autoplay"></iframe>

Try It

Kitten TTS Nano
Not loaded(24.0 MB)
0 / 5000

Try an example

Recent generations

No generations yet. Generate speech above to see your history here.

Technical Details

Architecture

StyleTTS2 variant

Parameters

15M

Model Size

24 MB

Voices

8

Languages

en

Links