This project is based on the ideas that our frame of mind is always influenced by our environment, and the intensity of light in a room has a strong influence on the atmosphere. So why not make the light act according to the atmosphere in a conversation?

I used an SVM algorithm, to train a model with tagged conversation sound samples, indicating how relaxed the conversation is.
The sound in the room is recorded all the time and sent to the trained model every 3 seconds for classification.
The light is then changed according to the how intimate or angry the conversation in the room is.