Music Recommendation

While working as a ballroom dance teacher, I noticed that there was no reliable app for identifying classical dance styles. Finding suitable music therefore required repeated manual searching and curation.

I built an AI system for music classification, inspired by current audio-analysis research. The system combines metadata from Spotify, a custom web scraper for YouTube Music and my own curated dance-music database for fine-grained labels.

The application identifies dance styles and generates network graphs to capture similarities between songs. This enables personalized playlist generation and recommendations that adapt to user behaviour over time.

The app is called Bastibeatz and is my personal daily music tool. Because of Spotify policy restrictions, I do not publish it publicly.

Technologies: Kotlin (Android), Python, SQL, Machine Learning