Abstract

Our project investigates environmental soundscapes in West Lafayette and on the Purdue campus - including the Purdue Airport with its newly reopened commercial terminal - treating these sites as a living laboratory. Conducted under the Discovery Undergraduate Interdisciplinary Research Internship (DUIRI) with mentorship from Dr. Tae Hong Park and his Citygram research, the study integrates engineering, data science, and acoustic ecology.

Building on the Soundscaper system, we are developing an interactive sound-mapping interface that allows users to hear synthetic soundscapes representing real-world conditions. We implement a geospatial-audio pipeline that links OpenStreetMap (OSM) data to our Object-Oriented Soundscape Synthesis (OS2) framework. OSM provides coordinates and local feature tags (roads, buildings, water, land types, airports, etc.) which are parsed into sound objects representing elements in a soundscape. These objects are transmitted over OSC to a real-time MATLAB engine, with the OS2 method mapping each feature to a synthesized sound analogue. This enables interactive, data directed soundscape rendering, supporting rapid prototyping of soundscape simulations across large geographic regions.

Soundscaper focuses on developing tools for "generative soundscape synthesis" through what we term Object-Oriented Soundscape Synthesis (OS2): to construct interactive sound maps that allow users to click on a location to hear its sounds. OS2 is grounded in a codebook of sound templates such as human voices, vehicular motion, rainfall, bell sounds, and layered dynamic ambient textures. By "sound-stitching" together these templates, the system emulates contextually and dynamically responsive, meta/data-driven soundscapes. OS2 supports scalability across large geographic regions, with the Purdue campus area serving as a testbed for prototyping.

Additionally, we use the Citygram and GetNoisy sound-sensor network systems to track airplane noise data, which are analyzed through signal processing, automatic classification, and sonification techniques. We utilize audio data sets from several major airports such as JFK, as well as local locations at Purdue to train our machine learning model. Once imported into MATLAB, the audio events are classified into several categories (flyover, non-airplane, etc.). Data is then further cleaned and used in model training to better help identify and differentiate aircraft sound data.

This research, contributing to the Citygram Project and the GetNoisy startup initiatives, advances smart sound-sensing networks for aircraft noise tracking. Integrating IoT-enabled sensors, edge computing, and AI, the work encompasses dataset development, AI modeling, and data analysis. Using Purdue’s newly reopened commercial airport as a living "lab," we explore the adaptability and scalability of these systems to airports and urban contexts worldwide.

We examine the spatial, temporal, and perceptual dynamics of environmental sound, emphasizing implications for environmental sustainability, community engagement, and public well-being. Noise from the Purdue Airport and other urban sources is interpreted as a living trace of community activity, informing scalable tools for interactive sound mapping, environmental monitoring, and civic engagement. By combining humanistic listening with computational analysis, this project treats cities as resonant systems in which every sound reveals stories of technology, ecology, and collective experience. The project advances soundscape-based approaches to understanding the environment while developing new tools for monitoring and engaging with the sonic life of communities.

Keywords

Noise Pollution, Sensor Networks, Urban Health, Acoustic Ecology, Object-Oriented Sound Synthesis

Date of this Version

2025

Download

Share

COinS