Beat Bot is a machine learning algorithm that determines the "mood" of a song. As it does this, it performs real time visualizations of what it believes would fit the mood. This is accomplished by using TensorFlow to reward the algorithm when it correctly identifies a sample of music as "calm" versus "energetic".
The goal is to create an application that can accurately categorize sections of music as a climax or a reprise through the intensity of its musical components. This can aid music producers and music theorists in their compositions and studies.
In the future, we plan to create more visualizations, generate data indicating the overall feeling of a song, and provide statistics a producer can use to improve their music. The algorithm itself can also be taken music streaming services as a form of curation and entertainment. Made in 24 hours at Hack The U.
It is a 4 person project where I handled most of the visualization and user feedback. This includes:
- Implementing a skybox manager to dynamically change its color based on the mood of the music
- Developing a mood bar to indicate what the energy the music is outputting
- Creating a VFX manager to spawn particle systems based on the music