Music exerts a powerful influence on the brain, with the potential to enhance health and alleviate various symptoms. As a core element of human experience, music has shaped cultures and periods throughout history. In the present day, music encompasses a vast array of types, blending contemporary styles with historical traditions. Each type is characterized by distinctive forms, expressions, and techniques. Individual music preferences vary significantly, and many listeners are often unaware of the types they favor. This study involved extracting sound features from music data and training a Keras model to classify these types, achieving an accuracy of 72.33% in type classification. The sound features were extracted using methods such as MFCC, Mel spectrogram, Chroma vector, and Tonnetz from the Librosa library. These features were then utilized with TensorFlow/Keras to classify popular songs identified by Shazam in Turkey. With numerous classification methods available, selecting the most appropriate one can be challenging. This study provides valuable insights into type classification using Keras and assists those unfamiliar with music in recognizing and categorizing newly released songs.