An Interactive Map
Suhan Kacholia, Franklin College of Arts and Sciences, Institute for Artificial Intelligence, Cognitive science, 4th year undergraduate
This project maps the musical history of Athens using archival interviews, large language models (LLMs), and geospatial visualization. The core source of data is the Athens Music Project Oral History Collection, maintained by the Richard B. Russell Library for Political Research and Studies and initiated in 2014. The collection consists of over 120 interviews that document various musical traditions in Athens, including the Southern independent rock scene, African-American musical traditions, hip-hop, jazz, bluegrass, folk music, Latin music, new music, conceptual sound art, classical music, and musical theater. The interviews also cover the development of AthFest, an annual three-day festival that supports music and arts education.
To build the map, I first processed the audio transcripts from these interviews using Google’s Gemini-2.0- Flash multimodal AI model. The AI takes the audio file and extracts specific location data by identifying the venues and landmarks mentioned in the interviews. The output includes the exact venue names along with a brief description of their significance in Athens’ musical history, the original interview title, and a link to the interview. This approach ensures that the information is directly connected to the archival source.
After the location extraction, I used the Python library geopy to perform geocoding. Geocoding converts the venue names into latitude and longitude coordinates, which can be plotted on a map. However, any location that no longer exists or lies outside the city limits is removed from the dataset. This step ensures that only relevant, geographically correct sites are included in the final output.
The geocoded data is then integrated with Mapbox to create an interactive web-based map. Each location appears as a marker on the map. When a user clicks a marker, a popup displays the interview title, the description of the venue, and a clickable link to the original interview. If multiple interviews reference the same location, the information is grouped together so that all related details are accessible in one popup.
I was inspired by local initiatives such as the Athens Music Walk of Fame and the Athens Music History Self Guided Tour that help situate the rich musical history of Athens in a spatial format. I hoped this project would help (1) demonstrate how AI can be used to transform large amounts of unstructured data into useful, human-interpretable insights, (2) help bring to life the Athens Music Project Oral History and UGA’s collections of archival research and (3) help students connect with the rich musical legacy of Athens and follow in the (literal) footsteps of past Athens creatives and musicians.