Özet:

People are constantly engaged in communication with each other, and they mostly do so through language. The most effective form of communication for a newborn baby until they acquire this skill is crying. Although baby cries are often perceived as bothersome by adult individuals, they can contain a wealth of information. In this study, our goal is to interpret the information embedded in baby cry audio signals using sound processing methods and classify them using machine learning algorithms. To achieve this objective, we utilized a dataset consisting of baby cry audio signals divided into five distinct classes. Feature extraction operations were applied to the dataset, and performance metrics were measured using classification algorithms. Subsequently, to examine the impact of data augmentation on performance metrics, the data was partitioned into equal segments. The changes in performance metrics were analyzed based on the applied data augmentation technique, and it was determined that the employed method enhanced the classification accuracy.

Anahtar Kelimeler: