top of page

Acoulatis user guide -
Accuracy of our prediction model

How accurate is the software, which margins should I keep in the calculations?

​

The accuracy in the predictions is typically within a 3 dB margin with higher accuracy for manufacturer products used together with massive elements, like a raised floor or a suspended ceiling. We also have a neural network AI prediction model for the lightweight frame walls with high accuracy. A paper about our neural network model was presented at DAS/DAGA in Copenhagen. You can read the paper here.

​

There are two papers available about the accuracy of the software and comparison between other prediction software on the market. You can read the one from Internoise here, and the one from Euronoise here.

​

​In the approximated field value, we include a margin before we present the weighted values depending on the selected junction type. We also have three different colours that indicates if the construction is able to fulfil the set requirement. Green means that there is a good margin to requirement. Yellow means that the construction could work but it needs to be investigated in more detail. Red means that the construction would most likely not work for the set requirements.

bottom of page