Hammering Sound Evaluation under Drone Noise Environment using Noise Reduction and Machine Learning Methods

Abstract

In this study, we considered a method to evaluate a hammering sound component using machine learning for construction inspection in drone noise condition. In the test, hammering inspection was performed using two tiles (defective and normal tiles) with or without drone noise. For carrying out automatic hammering sound evaluation, machine learning model was prepared using a lot of hammering sounds obtained at quiet condition. Then we applied the model to the hammering sound with drone noise to evaluate the tile conditions. However, the accuracy was insufficient due to the noise. For the improvement of the accuracy, we made machine learning model again using standardized SPL in each 1/3 octave band to reduce the noise influence. In addition, we also applied a noise reduction method using acoustic transfer function. As the result, the accuracy was improved so much and the automatic hammering sound inspection method under drone noise condition could be realized.