This Might be the World’s First AI-Enhanced Talking C-3PO Head

Samuel Potozkin spent months shaping a life-size C-3PO head from plastic filament and careful layers of paint. What started as a simple Star Wars fan project grew into a working machine that listens to spoken words and replies in the droid’s familiar voice. Anyone who has ever wished the golden protocol droid from Star Wars could answer back now has a real version sitting on a shelf or desk.

Potozkin printed the head on a Prusa Core 1 machine with PETG filament and a few wobbly layers. Hours of hand sanding later, the curves are velvety smooth. Primer was then used, followed by many layers of Alclad II chrome paint to achieve the desired reflecting gold surface. To achieve that movie-accurate gloss, apply a final layer of yellow-orange color and seal with clear 2K resin. There was a hiccup when the object fell off the desk and a crack formed, but a dab of glue and putty fixed it and work resumed as usual.

Inside the hollow head, there’s a Raspberry Pi 5, a tiny microphone, and a speaker, all neatly wired. The Pi runs a special piece of programming that manages the conversation. The sound from the microphone is routed through a Whisper-based speech-to-text system and then to a large language model, which generates a response in the style of C-3PO, the bot from the films. A voice-synthesis layer then converts the text into proper speech that sounds similar to Anthony Daniels’ original performance, with a speaker pushing it all out so the head appears to be speaking to you.

Conversations with the head feel natural, even if there is a small delay between asking and receiving an answer. Ask who the droid is, and you’ll get a nice protocol-droid response. When you ask about Earth, it regales you with statistics in the same super-formal tone that fans associate with the films. The lag was a bit of an issue at first, but with a few code modifications, everything is fine again. When it comes to running the device, no further equipment is required because it is completely self-contained and does not require a separate computer or smartphone app.

He’s also made the complete code and 3D files available on GitHub so that anyone can try their hand at creating their own version. If you’re curious about how the software pipeline works, there’s a lengthy paper on the same repository that explains everything. Potozkin believes that this effort is simply a modest step toward machines that can sit with humans in the real world rather than merely live on a screen all the time.
[Source]?