Abstract
Automatic urban sound classification is a desirable capability for urban monitoring systems, allowing real-time monitoring of urban environments and recognition of events. Current embedded systems provide enough computational power to perform real-time urban audio recognition. Using such devices for the edge computation when acting as nodes of Wireless Sensor Networks (WSN) drastically alleviates the required bandwidth consumption. In this paper, we evaluate classical Machine Learning (ML) techniques for urban sound classification on embedded devices with respect to accuracy and execution time. This evaluation provides a real estimation of what can be expected when performing urban sound classification on such constrained devices. In addition, a cascade approach is also proposed to combine ML techniques by exploiting embedded characteristics such as pipeline or multi-thread execution present in current embedded devices. The accuracy of this approach is similar to the traditional solutions, but provides in addition more flexibility to prioritize accuracy or timing.
Highlights
Despite the potential application of urban sound recognition in Wireless Acoustic Sensor Networks (WASN), there exists a lack of evaluation of existing solutions
The best performing classifiers are evaluated on an embedded system in order to determine the achievable accuracy and the execution time that could be expected on such constrained devices
Our analysis provides valuable information about the achievable accuracy and the required execution time on embedded devices
Summary
Despite the potential application of urban sound recognition in Wireless Acoustic Sensor Networks (WASN), there exists a lack of evaluation of existing solutions. The achievable classifier’s accuracy or the demanded time to perform the sound classification are just a couple of parameters, which must be taken into account when considering WASN applications demanding sound recognition. Current embedded systems provide enough computational power to perform urban noise classification, enabling edge-computing solutions. The presented work evaluates the most popular Machine Learning (ML) techniques for Environment Sound Recognition (ESR), in particular for urban sound classification. Existing open source libraries for audio analysis and datasets for ESR are used to evaluate the achievable accuracy of classical sound classifiers. The best performing classifiers are evaluated on an embedded system in order to determine the achievable accuracy and the execution time that could be expected on such constrained devices. A scalable approach is proposed to exploit some characteristics of embedded devices and to enable the
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
