Here are some videos of it in action:
The way it works is I have an iOS shortcut that I press. It brings up the camera and takes a photo. iOS reads the label and pulls the text from it and then sends that to a sensor in Home Assistant. Once the sensor changes in Home Assistant, it triggers an automation that uses the the conversation process service to give me information on the varietal of the grape, the producer of the wine, food pairings and any professional ratings. And a joke or two.
It also figures out if it’s a red, white or rose wine, and it changes the light on the 3D printed pedestal and behind the buffet accordingly. Then it uses a clone of Stephen Fry’s voice that I made with ElevenLabs to give me all the information. It also chooses from 10 random “AI Sommelier” videos I put together and plays them on the screen, and chooses from 25 classical type music clips I edited.
It also uses DALLE-3 to create a series of wine related images based on the information provided by the output from the conversation process service. One of a vineyard based on the location the wine was made, one of people picking the grapes, one of them pressing and bottling them, and one of a sommelier sampling the wine in a classy tasting room (with my cat somewhere in the image).
I had to get this done for a wine and cheese tasting party my neighbor won in a charity auction which we’ll be having at my house next weekend. But I’m planning on adding some more features. At the minimum a switch that can tell when a bottle is placed or removed from the 3D printed pedestal. I’d also like to see if I can use an ESP32 Cam with a wide angle lens coming off of the front on a low arm to take a photo of the bottle label and pass it to the Extended OpenAI Query image service, since that does a better job of figuring out the exact wine. The current way I do it with iOS extracting the text isn’t completely reliable.
If anyone is interested and has more specific questions, let me know!