Using the GitHub link given below, you can explore the modular components that generate context-aware, humorous poetry.

A photo is captured with the Raspberry Pi Camera, recording the user’s expression, posture, and surroundings. The image is sent to LLaVA-13B, which generates a scene description, and then to DeepSeek V3 to create an 8-line free verse poem. Finally, Kokoro-82m converts the text into a human-like spoken voice, delivering a witty, spoken poem that humorously reflects the user and their environment.

Github Repository