Raspberry Pi smart home technology just keeps getting more fun and easier to access, thanks to neat projects like this one from maker DevMiser. Their latest creation, known as Lumina, brings AI image generation into your living room. It uses a Raspberry Pi to create artwork for you on the fly using open-source AI tools.
The device is designed to plug into a TV via HDMI. Instead of using a keyboard for input, it relies on speech-to-text to retrieve prompts for the AI image generator. Just tell Lumina what sort of image you want to be created and wait for the request to be processed. Once it's created an image, it will display automatically on the TV.
According to maker DevMiser, Lumina uses a custom Python script. It calls upon a few other tools ,including Picovoice, to handle the speech-to-text, as well as DALL-E 3 from OpenAI to generate the images. All of the software is handled by a Raspberry Pi 4, but you could definitely get away with using a Raspberry Pi 5 if you've got one on hand.
The Raspberry Pi is connected to a USB microphone that constantly monitors speech, listening for the right trigger word to initiate the image request process. All of the hardware is housed inside of a custom 3D-printed case. You can find the STL files for the case, as well as all of the source code for the project over at GitHub.
DevMiser was kind enough to make Lumina an open source project, and provides detailed instructions for anyone who wants to recreate this Raspberry Pi project at home. The official project page over at GitHub walks you through setting up the hardware, as well as how to use both Picovoice and OpenAI to make the voice commands work.
