Melody of Budapest by Magyar Telekom Nyrt.

The more we observe a city, the more apparent it becomes that it surpasses the sum of its parts. It resembles a colossal, intricate organism with its network of organs and unique internal dynamics. These dynamics are a tapestry woven from our daily interactions within the city: the bustling morning and evening rushes on the tram or subway lines, leisurely strolls along the riverbank — it all contributes to this vibrant tapestry.

Though it sounds idyllic, it stands worlds apart from the vast data streams of corporations, the omnipresent data visualization tools, the ongoing buzz about artificial intelligence across various platforms, or even the fundamental essence of artistic expression — music.

"Melody of Budapest," sponsored by Telekom Hungary, aims to bridge these divides by transforming the city's dynamic mobility patterns into a musical piece using data science and AI. Initially, Janosov and colleagues processed large-scale cell tower-based mobility data into digestible information. This unpublished data covers the temporal and spatial specifics of every mobile phone event on Telekom's network in Budapest, Hungary, from June 2022 to 2023.

Next, the aggregated mobility intensity across the entire city was converted into a time series depicting varying day and night activity levels. This series was then translated into machine-produced audio, converting each intensity data point into a musical note in C major. Thus, a day equates to two data points, resulting in roughly 730 musical notes per year.

Subsequently, Janosov collaborated with the generative AI composer Aiva, trained on a substantial library of classical music, to craft the human-AI sound combination presented in the video. Completing the "Melody of Budapest" entailed visual representation that mirrors the city's vibrant pulse.

To achieve this, a three-dimensional, audio-reactive map visualization was created, showcasing an entire year in Budapest. This visual divides the city into its 23 districts, revealing each district's relative intensity of cell activity throughout day and night for one year. These intensity levels vividly outline each district's characteristics — the lively city center, serene suburbs, daily mobility patterns, and the city's nocturnal and weekend rhythms.

Lastly, it's worth highlighting the three layers through which these visuals convey Budapest's underlying dynamics. The columns directly display normalized district-level activity. The audio line is derived from mobility time series data, augmented by AI. The animation's coloring responds to changes in music intensity, complemented by subtle random noise elements. This multi-layered audiovisual product offers an immersive experience, visualizing the "Melody of Budapest" based on a large-scale underlying database.