gradium-ai/xn-ptts ? reverse-engineered prompt

Build me a fast offline text to speech app called Phonon that can turn text into a WAV file from the command line, and also work as a reusable speech engine for Python, browsers, iPhone and Mac apps, and Android. It should support English, French, German, Spanish, and Portuguese, let me choose a voice, and let me adjust how fast it speaks. I want it to stream audio as soon as possible instead of waiting for the whole sentence, and it should run well without needing PyTorch or a cloud service. Please make it easy to install and use on desktop, mobile, and in the browser, and include simple examples so I can test it right away. If you need to check current docs for any platform specific setup, do that.