Speaker Recognition Based on Machine Learning Classifiers for Parent and Child Speaker Diarization

In this study, we present speaker recognition methods based on multi-layer perceptron (MLP), support vector machine (SVM), random forest (RF), and K-nearest neighbor (KNN). By standardizing MFCC features and slicing speech data using sliding window technology, we adjusted the model parameters to improve classification accuracy. Our experimental results showed that these methods performed well in identifying speakers, especially when the intonation, pitch, and speech patterns of parents and children were similar. We hope that this study will provide a valuable reference for future speaker diarization research.