With the increasing emphasis on machinery reliability and maintenance optimization, early fault diagnosis in mechanical systems such as gearboxes has become critical. This paper presents a comprehensive review of both traditional machine learning (ML) and deep learning (DL) techniques for gearbox fault diagnosis and anomaly detection. We systematically collect data from various gearboxes under different operational conditions and compared the different methods for fault diagnosis scenarios. State-of-the-art Machine Learning algorithms are compared against deep learning architectures. In this paper, we have presented a review of various ML and DL techniques suggested for gearbox fault diagnosis. We have also discussed the challenges of the existing techniques. Through this comparative study, we provide a roadmap for practitioners and researchers to select the most suitable technique for their specific gearbox fault diagnosis needs, thereby contributing to the efficient maintenance and prolonged service life of numerous gear boxes.