This thesis explores the integration of quaternion algebra into neural network architectures to enhance their efficiency for diverse audio processing tasks. Quaternion-based transformations are employed to achieve structural ...