ABSTRACT

My research represents an innovative approach that is able to classify gender and singers in duet compositions using NN. Utilizing a multi-layer perceptron (MLP) neural network, the research delves into the intricate process of classifying gender and singers from a dataset of duet songs. The methodology begins with the collection of a diverse range of duet songs from various sources, ensuring a broad representation of genders, genres, and vocal styles. These songs undergo a meticulous preprocessing phase, including noise reduction, normalization, and segmentation, to prepare clean and uniform audio data for feature extraction. Key characteristics for example are MFCCs, Chroma Shape, Contrast Shape, and others are extracted and normalized to facilitate efficient NN training.

The study then proceeds to train the MLP neural network using these extracted features, concentrating on fine-tuning hyperparameters for precise classification. The neural network's performance is rigorously assessed through testing, validation, and metrics such as accuracy, precision, recall, and Fl-score. Results show that the neural network model effectively determines gender and singer identity in duet compositions with notable precision and few errors. This research highlights the capabilities of neural networks in music information retrieval and sets the stage for future developments in more sophisticated systems like neuro-fuzzy neural networks, which classification outcomes. could offer improved clarity in