Processing and Recognition of Pressure Cooker Whistle Sound using MFCC and K-NN

A pressure cooker has made it possible for the survival that cooking can be done in a matter of few minutes. A weight is placed on a pressure cooker and every time the pressure rises above the threshold pressure in the sealed chamber, the weight creates a whistle sound. People use this whistle as an indication that their food has been cooked. But they often tend to forget to keep track every time it is raised. When left unattended for a long time, it might lead to flaring up of the cooker, leading to explosions. Few approaches have addressed this problem but weren't effective in Realtime. Hence our approach is to process and recognize the sound the whistle makes every time it is raised, using MFCC and K-NN. Since there was no proper availability of dataset, datasets were created. Audio samples about 62 were used to build the machine learning model and accuracy of 88.6% was attained.