What is the feature os this project?

Hi, thank you for the new job for TTS, is there any paper to introduce the job? I do not know the idea of this project, is it targeted at to producing speech from text with the same quality like token based LLM-TTS and with fewer clean training data and very fast inference? or it is a e2e tts more natural then FastSpeech2/VITS like model but can not reach the quality of LLM-TTS? thank you.