This project uses llama.cpp as an LLM server to perform inference and generate speech using Synthetic voice library - ayutaz/uni-llm-voice-chat