When the engine of a motor-cycle goes into its defective stage, it affects the performance of the motorcycle and hence decreases its efficiency. Therefore, for the smooth running of the motorcycle, the health monitoring of its engine is very essential which increases its life and efficiency. Acoustic emission signature from the motorcycle engine is a measuring parameter to identify the faults in the engine. During its healthy stage, the motor vehicle engines generate a specific acoustic emission signature. But when the engine goes into the defective stage then a significant change in these acoustic emission signature occurs. This change can be used to diagnose the defects in motor engines. The objective of this work is to diagnose the faults in the motorcycle engine using statistical and time-frequency analysis of the acoustic emission signal and then classify the faults using machine learning classifiers. In this experimental work, an instrumentation system is used to acquire the acoustic emission signal from the engines under different conditions. Then the statistical parameters from the acoustic signatures are computed and compared to identify the faults. The wavelet analysis of the acquired acoustic emission signal is also done to diagnose the faults. The classifiers are used to classify the faults by using the statistical parameters as its input. At first stage, the experimental set-up is developed to acquire acoustic emission signatures from the healthy and defective engines. In second stage, signal analysis using statistical and wavelet signal processing technique is done to identify the faults in the engine and in the third stage, the faults in the engine are classified using the machine learning techniques. The result of the proposed work shows that the classification accuracy of the random forest classifier is better than the decision tree classifier. The novelty of the proposed work is that the wavelet analysis along with machine learning technique is used to diagnose the faults in the engine