Synthesize speech right here — WebGPU + WebAssembly, nothing leaves your machine. First load downloads the model weights (cached after).

Pick an engine and load it.

Text to synthesize

Output