News

Check out our new video describing the SONYC project!

Our paper “Feature Learning with Deep Scattering for Urban Sound Analysis” has been accepted for publication at the 2015 European Signal Processing Conference.

In this paper we evaluate the scattering transform as an alternative signal representation to the mel-spectrogram in the context of unsupervised feature learning for urban sound classification. We show that we can obtain comparable (or better) performance using the scattering transform whilst reducing both the amount of training data required for feature learning and the size of the learned codebook by an order of magnitude. In both cases the improvement is attributed to the local phase invariance of the representation. We also observe improved classification of sources in the background of the auditory scene, a result that provides further support for the importance of temporal modulation in sound segregation.

The full paper is available on the Publications page.

Our paper “The design and calibration of low cost urban acoustic sensing devices”, presented at Euronoise 2015 in Maastricht, was awarded the best paper and presentation award with an award of €500!

The presentation can be viewed here and the full paper is available on the Publications page.

SONYC team member Charlie Mydlarz was invited to attend the NSF Early-Career Investigators Workshop on Cyber-Physical Systems in Smart Cities in Seattle, April 13-17, 2015, with a travel stipend award of $1500 based on the submitted Position Paper.

SONYC was also offered the opportunity to present to the workshop attendees including a poster session after the talk with an extremely positive response.

Our paper “Unsupervised Feature Learning for Urban Sound Classification” has been accepted for publication at the 2015 International Conference on Acoustics, Speech and Signal Processing (ICASSP).

Recent studies have demonstrated the potential of unsupervised feature learning for sound classification. In this paper we further explore the application of the spherical k-means algorithm for feature learning from audio signals, here in the domain of urban sound classification. Spherical k-means is a relatively simple technique that has recently been shown to be competitive with other more complex and time consuming approaches. We study how different parts of the processing pipeline influence performance, taking into account the specificities of the urban sonic environment. We evaluate our approach on the largest public dataset of urban sound sources available for research, and compare it to a baseline system based on MFCCs. We show that feature learning can outperform the baseline approach by configuring it to capture the temporal dynamics of urban sources. The results are complemented with error analysis and some proposals for future research.

The full paper is available on the Publications page.

Our paper “The design and calibration of low cost urban acoustic sensing devices” has been accepted for the 10th European Congress and Exposition on Noise Control Engineering (EuroNoise) in Maastricht, The Netherlands, May 2015.

This paper details the sensors’ design, development and potential future applications, with a focus on the calibration of the devices’ Microelectromechanical systems (MEMS) microphone in order to generate reliable decibel levels at the type/class 2 level.

The full paper is available on the Publications page.

Our paper “A Dataset and Taxonomy for Urban Sound Research” has been accepted for publication at the 22nd ACM International Conference on Multimedia (ACM-MM’14).

Automatic urban sound classification is a growing area of research with applications in multimedia retrieval and urban informatics. In this paper we identify two main barriers to research in this area – the lack of a common taxonomy and the scarceness of large, real-world, annotated data. To address these issues we present a taxonomy of urban sounds and a new dataset, UrbanSound, containing 27 hours of au- dio with 18.5 hours of annotated sound event occurrences across 10 sound classes. The challenges presented by the new dataset are studied through a series of experiments using a baseline classification system.

The full paper is available on the Publications page. You may also wish to visit the UrbanSound dataset companion website.

Our paper “The design of urban sound monitoring devices” has been accepted for the Audio Engineering Society’s 139th Convention in Los Angeles, October 2014.

The paper describes the design and build of a self-contained urban acoustic sensing device to capture, analyze, and transmit high quality sound from any given urban environment. The presented acoustic sensing device prototype incorporates a quad core Android based mini PC with Wi-Fi capabilities, a custom MEMS microphone and a USB audio device. The design considerations, materials used, noise mitigation strategies and the associated measurements are detailed in the following paper.

The full paper is available on the Publications page.

Our paper “The implementation of MEMS microphones for urban sound sensing” has been accepted for the Audio Engineering Society’s 139th Convention in Los Angeles, October 2014.

The paper details the microphone selection process, involving the comparison between a range of consumer and custom made MEMS microphone solutions in terms of their environmental durability, frequency response, dynamic range and directivity. Ultimately a MEMS solution is proposed based on its superior resilience to varying environmental conditions and preferred acoustic characteristics.

The full paper is available on the Publications page.

Last weekend SONYC researchers Charlie Mydlarz and Justin Salamon participated in the Science in the City hackday which took place as part of the NYC 2014 World Science Festival. Their project, a web-app for crowdsourcing urban sound tagging, won the “Best Social Impact” award.