Detection of turkish commands for chess game control with artificial intelligence

Within the scope of the thesis, a chess-playing study is being developed that individuals with limited mobility can use in cafes and social centers created for disabled individuals. It is aimed to carry out a study that will help people who were born without limbs and lost their limbs due to a certain accident or disease, so that they can continue their social lives even if they do not have a chance to be treated. With this study, it is aimed to help people with disabilities in their rehabilitation as well as their socialization. It consists of three structures: chess module, human-computer interaction module and artificial intelligence module. These structures, which work separately, have been transformed into a system that works together. A system was developed using 43790 voice recordings taken from 151 people, 86 men and 65 women. Mel frequency kepstral coefficients (MFCC) and gammatone kepstral coefficients (GTCC) method were used as feature extraction method from sound recordings. Maximum normalization was applied to these extracted features. In addition to classification processes with k nearest neighbor (k-NN), support vector machine (SVM), naive bayes and multilayer networks (ÇKA), the extracted features are classified using convolutional neural networks (CNN) model. Data sets obtained by MFCC and GTCC methods were used in the CNN model. More than 85% success has been achieved and 93% success has been achieved with person-based results.