Audio8-AI/Audio8_TTS ? reverse-engineered prompt

Build me a small Python app for text to speech that can also clone a voice from a short reference recording.

I want to type some text, optionally give it a sample audio clip and transcript, and get back a WAV file in the same voice. It should support batch runs from a simple JSONL file too, so I can generate a folder of clips from a list of prompts. Please make it work with the provided model checkpoint, and keep the setup simple with clear install and run commands.

If it helps, include a local way to test it from the command line, and a lightweight CPU friendly option if available. It should work best for the supported languages in the repo, and if the text is too long, split it or warn me so the output stays good. Look up current docs online if you need to.