Sound detection for study room monitoring and evaluation

In the field of sound recognition, the research and study of sound event detection are still active, with the vast majority of the papers focusing mostly on the domain of speech and music. This paper presents and discusses a framework for a study room event detection system. Feature extraction techniques are utilized and discussed to obtain the parametric type representation for the analysis of the sound for intelligent homes machine listening systems specifically for study room. The conduct of sound analysis within the category of the sounds the least accurate was the door knock, but the accuracy of 95.00%, currently in the field is acknowledged as good, making the parameters fit for detecting surrounding sounds. The performance of the CNN in detecting environmental sounds was analyzed using the parameters that were defined, with an overall accuracy of 96.8%. The result was promising for machine learning that detects sounds that can be applied as technology for an innovative learning environment.