Text to speech, streamed from your own CPU

Phonon 31M compiled to WebAssembly with fusor-ai. It runs on your CPU, streams every 80 ms frame to your speakers as it is computed, and once loaded, nothing you type leaves the page.