Audio-based pedestrian detection is a challenging task and has, thus far, only been explored in noise-limited environments. We present a new dataset, results, and a detailed analysis of the state-of-the-art in audio-based pedestrian detection in the presence of vehicular noise. In our study, we conduct three analyses: (i) cross-dataset evaluation between noisy and noise-limited environments, (ii) an assessment of the impact of noisy data on model performance, highlighting the influence of acoustic context, and (iii) an evaluation of the model's predictive robustness on out-of-domain sounds. The new dataset is a comprehensive 1321-hour roadside dataset. It incorporates traffic-rich soundscapes. Each recording includes 16kHz audio synchronized with frame-level pedestrian annotations and 1fps video thumbnails.

Noise is one of the primary quality-of-life issues in urban environments. While low-cost sensors can be deployed to monitor ambient noise levels at high temporal resolutions, the amount of data they produce and the complexity of these data pose significant analytical challenges. One way to address these challenges is through machine listening techniques, which are used to extract features in attempts to classify the source of noise and understand temporal patterns of a city's noise situation. However, the overwhelming number of noise sources in the urban environment and the scarcity of labeled data makes it nearly impossible to create classification models with large enough vocabularies that capture the true dynamism of urban soundscapes In this paper, we first identify a set of requirements in the yet unexplored domain of urban soundscape exploration. To satisfy the requirements and tackle the identified challenges, we propose Urban Rhapsody, a framework that combines state-of-the-art audio representation, machine learning, and visual analytics to allow users to interactively create classification models, understand noise patterns of a city, and quickly retrieve and label audio excerpts in order to create a large high-precision annotated database of urban sound recordings.

What happens when the sensor-imbued city acquires the ability to see – almost as if it had eyes? Ahead of the 2019 Shenzhen Biennale of Urbanism\Architecture (UABB), titled "Urban Interactions," ArchDaily is working with the curators of the "Eyes of the City" section at the Biennial to stimulate a discussion on how new technologies – and Artificial Intelligence in particular – might impact architecture and urban life. Here you can read the "Eyes of the City" curatorial statement by Carlo Ratti, the Politecnico di Torino and SCUT. Technologies of the virtual realm present an opportunity to rethink the experience of space, society, and culture. They give us the possibility to engage with the city of the future, shaping the built environment of the 21st century.

Further, recent psychoacoustical evidence suggest the approach bears some resemblance with human auditory processing for sound textures (McDermott et al., 2013; Nelken and de Cheveigné, 2013). In an influential 2007 article, Aucouturier, Defreville & Pachet (Aucouturier et al., 2007) applied a BOF model to categorize both polyphonic music and soundscapes. Their results showed that, while BOF was a meriting model for their polyphonic music dataset, it was spectacularly effective for soundscapes, reaching accuracies of 96%. The contrast, they interpreted, lied in differences in the temporal structure of both types of stimuli, with music being more formally organized and soundscapes more easily summarized by statistics. In a later companion study (Aucouturier and Defreville, 2009), they showed that soundscapes could be time-shuffled without altering listeners' perception of their acoustic similarity, while music could not. While more work was needed for music, the authors therefore concluded that BOF was a sufficient model to approximate human perception for soundscapes, practically ruling out the need to recognize the local acoustic events in a texture in order to identify it.