Smart Dance Generator

The Project

I used python and dart to create an app that can choreograph dance moves, displaying a stringed together video of stick figure images, creating a stop-motion effect. The app takes in a short clip of music, and by analyzing each note and comparing it to the songs in the database, the top three most similar songs are chosen. Out of these three songs, clips are taken from choreography already given to the songs (from the database), and the dance videos are overlined with a motion tracker to trace out the image of a stick figure. These images are then concatenated and combined to create the stop-motion video choreography for the user-inputted audio. By taking into account the music, choreography is aligned with the musicality. This app works as a way for dancers and choreographers to gain inspiration and be able to focus on enjoying their craft. I hope that my app can be used to successfully eliminate stress from a pressure-relieving activity and offer motivation for dancers struggling with crafting authentic choreography.

View this project on YouTube

Team Comments

I currently dance on my high school dance team, and I have been dancing for 14 years. Something that my peers and coaches struggle with is dancer's block. Similar to writer's block, oftentimes it is hard to create original and creative choreography. My app choreographs according to the given audio.

The first challenge was figuring out how to analyze music and compare the songs to find the most similar songs to the inputted audio in the database. The librosa library was used to compare individual notes and the similarity levels between notes were used to rank songs in order of their similarity.

With more time I would develop a subscription feature which would allow the user to receive longer clips of choreographed moves, since the current limitation is 30 seconds of choreography.