I Gave My ESP32 a Voice. Now It Judges Me.

I Gave My ESP32 a Voice. Now It Judges Me.

More

Summary

Creator Magic builds Pip, an AI-powered Tamagotchi-style pet running on a tiny ESP32 device with a camera, gyroscope, touchscreen, speaker, and motion sensor. Pip reacts to touch and shaking, reads the user’s mood through its camera, and can speak about things like YouTube channel stats.

The entire project is vibe coded with Claude Code, without manuals. The builder sets up API keys for Fish Audio, Claude, and YouTube, then asks Claude to create a local audition site for comparing voices. The video highlights Fish Audio’s new Drama 3 model, which performs lines with direction such as whispering or gasping rather than simply reading them. Side-by-side tests of flat and emotive voices show how much expression changes the character. The sponsor’s pay-as-you-go pricing is also discussed, with a $10 top-up proving ample.

From there, the build proceeds in steps: drawing the mascot face and animations, wiring up touch and motion reactions, adding speech with emotional delivery, connecting Wi-Fi and a Claude-powered brain, and finally enabling the camera. The creator plans to release the project as a free, open-source GitHub repository. It is a fun look at how AI coding tools and expressive speech models can turn inexpensive hardware into an interactive companion.


📺 Source: Creator Magic · Published October 23, 2026
🏷️ Format: Hands On Build

1 Item

Channels