suno-ai/bark ? reverse-engineered prompt

Build me a simple local text to audio tool based on Bark. I want to paste in a prompt and get back a wav file, with good defaults so it works out of the box. It should handle normal speech, different languages when possible, and fun things like laughter, singing style prompts, or simple background sounds if the model supports them.

Please include an easy way to choose from the built in voice presets, and make sure longer text is handled by splitting it into sensible chunks so the voice stays fairly consistent. I would also like a basic command line flow where I can pass text and an output filename, plus a small example notebook or demo script that shows the main features.

Set up the project so the models preload properly, save audio at the right sample rate, and document how to run it on either CPU or GPU. Keep it practical and easy to try locally. If anything is unclear, check the current Bark docs and examples online.