Automatic Classification of Sounds Using Feature Extraction and A.I.
Large European cities have Sound Pressure Level (SPL) monitoring stations that report the ambiance dBA level, and create average accumulative (Leq) levels of the sound, that the citizens are exposed to.
However, The authors have found that a one minute average Leq (1min) of 65 dBA can be generated from a pneumatic hammer or from the sound tree branches with some traffic background noise. That fact illustrates how bad is ubiquitous Leq providing an abstraction of urban sounds.
This white paper describes new models to measure, characterize, interpret and represent urban sounds using audio, color and image using the FFT transform.
We make a proposal to use Artificial Intelligence tools to classify the sound at the sensor edge.
Finally this paper describes innovative models to evaluate urban sounds far beyond the measure of the Sound Pressure Level at the data platform side.
Click on the link to download the Automatic Classification of Sounds Using Feature Extraction and A.I.
