One of the great uses of Raspberry Pi is helping you build home assistants that can enhance your life in many ways. The latest one that caught our attention is DevMiser's open-source project called Lumina, which uses Raspberry Pi 4 to generate AI artwork and automatically displays it on TVs.

No, you don't need to attach a keyboard to give instructions about what image it needs to display on the TV. Instead, Lumina gives you the liberty to use your voice to depict the image and get the output in a matter of seconds. However, you'll need a Raspberry Pi 4 (earlier versions are not supported) with the legacy 64-bit OS to develop this DIY project. For converting voice requests to images, it'll use OpenAI's DALL-E 3.

Besides Raspberry Pi 4, you'll need a few more pieces of hardware, including a micro HDMI to HDMI cable for connecting Lumina to the TV, a USB microphone for talking to Lumina, and a 5v power supply. Additionally, you'll have to have OpenAI and Picovoice accounts to be able to obtain your secret OpenAI API key and secret Picovoice access key, respectively. The access key ensures that Lumina listens to your voice instructions, while the API key is needed for generating and displaying AI art images on your TV.

You can say "Lumina" for the device to start listening to your instructions. Try something like "Lumina, a cat wearing a hat sleeping on a pile of books." As per the demo video posted by DevMiser, you get the AI art in no longer than a couple of seconds. However, if your image specifications are long and complex, there might be a delay of a few seconds. That's because longer prompts require more computation and more processing power. But before even getting there, you'll need to prepare the Raspberry Pi 4 by following some easy step-by-step instructions given on the Lumina GitHub page.