Starting with rigging in C4D, the rigged object (mango in this case) is transported into Unreal Engine, tracks a human face, and shows the same human expression on it. After that, the speech-to-text and word capture functions will help recognize certain words during speech. Nest, it come up with word-related changes to the virtual character based on the algorithm. However, considering the limited timeframe, I switched to Spark AR to create a rough prototype to see what it looks like before getting deep into Unreal Engine. In spark AR, I made filters with face distortion, floating objects, face masks, and voice distortion. Besides, I also created a sample podcast content talking about "trends, social belong, and stanley cups" that I will continue developing to test the tech part.