NoiseScout utilizes AI-based noise classification techniques to automatically categorize different types of noise, allowing users to quickly and easily identify sources of excessive noise and take appropriate actions. The system can distinguish between different classes of noise, such as speech, machinery, traffic, and others, providing valuable insights for noise management.
NoiseScout Example: Noise Classification with AI
How it works
NoiseScout classifies different noises with scores that indicate how reliably they can be assigned to one of the defined categories. The score is not simply a percentage number. There is a rather complex algorithm used to determine a score. It involves a pre-trained (pre-loaded) system with a set of classifications. The best matches are then found, and the score is calculated based on how many of the different categories are recognized in the 10-second wave file.
A low score by no means indicates that the result is not accurate, rather that there are other classifications that matched, and while they not score as high, they did reduce the other scores. Also, there are sometimes super-classifications that score higher.
In the following example, ...
Vehicle |
Rail
Transport |
Train |
60 |
25 |
24 |
... the AI may classify a specific sound as being a Train, while, at the same time, classifying it with a higher score as being a form of Rail transport (which a train certainly is), and finally more generally, and with the highest score, as a type of Vehicle (which rail transport certainly is). In this example we can deduce that the sound was created by a train.
Using AI in the NoiseScout system helps streamline the noise measurement and analysis process, making it more efficient and effective.