Noise source classification
Novia™ is a machine learning algorithm that tells if a noise is generated by a work site, or not.
Novia is a value added technology to Sixense's noise monitoring services. When an alarm is triggered, the recorded sound wave is analysed and compared to a library of previously classified spectrograms. The outcome is the identification and classification of the noise sources with a level of reliability usually greater that 97%.
Novia is applied to a continuous and real-time noise monitoring station that is installed adjacent the site to be analysed.
When an alarm is triggered, the system analyses few seconds of the sound wave that has generated an exceedance.
The resulting spectrogram - or fingerprint of the sound - is compared to large library of previously recorded and classified signatures.
Sixense currently holds more than 1,000 different reference noise signatures and counting.
The outcome of this process is a real-time classification of the source that has generated the alarm. Is it linked to the construction site or urban environment?
Advanced classification may also identify the nature of the source; crane, jack hammer, pilling, excavator, etc... for construction sites - or sirene, helicopter, motorbike, music, etc.. for urban environments.
In addition, the sound wave is available as an audio file for playback and further investigations or verifications.