A room-scale experimental project to visualize the sound using lights. Pitch is mapped to color, and volume to the number of lights turned on. As such, the room decorated with the lights changes color in accordance with the sound of activities inside it (e.g. playing music, conversation)
I led a team of 5 people to do this project, and wrote most of the harder parts of the code myself (including the sound processing and socket server to stream the processed sound from a laptop to the Raspberry Pi)
The problem with this project is that audio readings tend to be very jarring, with discontinuous bursts of sound followed by silence very rapidly. Because of this, the pitch and volume changed dramatically far too quickly, and would have led to drastic, jarring changes in the lighting. To account for this, we implemented moving averages in order to smoothen out the transitions.
Due to limitations of the Raspberry Pi (it can't simultaneously utilize its sound card and control lights), we're temporarily using our laptop mics to process the audio. Due to this, I designed all the code regarding sound processing to be completely independent of the light controls, and to make sure the two components (laptop and Raspberry Pi) knew as little about each other as possible. Thus, the laptop streamed only pre-smoothened values of pitch and volume to the Raspberry Pi. I used the websocket protocol over TCP/IP to avoid latency issues.
I'm going to improve this after I'm done with school, just because it's a cool thing to have in my room. Here are my planned changes: