Hiroshiba/hihobot-synthesis — reverse-engineered prompt

Build me a small Python project that lets me synthesize speech in my own voice from a trained model. I do not need you to create the training pipeline from scratch, I will train the voice model separately using the nnmnkwii tutorial, but I want this repo to handle the inference side cleanly.

Please set up the package, installable environment, and a simple sample script that I can run to generate audio from text with my trained files. Make it easy to understand where the model and any related data should go, and make the sample actually show the full flow from loading the model to saving synthesized speech. I want the project to feel usable by someone who is not deeply technical, so keep the structure simple and add a short README with run instructions.

If there are missing details, look up the current docs online and make reasonable choices. I mainly want something I can clone, install, point at my model, and run.