This project is a text-to-speech model developed by Fish Audio, based on VQGAN+Llama technology. It can be easily run and fine-tuned on personal devices with only 4GB of video memory (16GB recommended), supports Chinese, English, and Japanese languages, and allows voice timbre adjustments, delivering excellent speech synthesis quality.