Die sinking electrical discharge machining is used extensively to produce very complex geometries on difficult-to-machine materials for aerospace, automotive, mold and die applications. The quality of the component produced mainly depends on the quality and intensity of spark generated between the tool and the workpiece. The present research work focused on using the acoustic emission signals to characterize the spark activity in the EDM process under different machining conditions. Experiments were conducted by varying the machining conditions and using three tool electrode materials namely graphite, copper and copper tungsten. The acoustic emission signal features were extracted from the data collected from different experimental conditions and are related to the various parameters including current, voltage, pulse on-time, surface roughness parameters (Ra and Rq) of the workpiece and material removal rate (MRR). The effect of tool material on the AE phenomenon was also investigated. The process conditions were found to have to have a good correlation with the acoustic emission signals features. Multiple linear regression models were built to predict the surface roughness (Ra) of the workpiece and MRR using only the current and pulse on-time as predictor variables. The prediction accuracy of the models improved by the addition of AE signal features to the predictor variable set. The results obtained indicate a suitability of employing acoustic emission signals for monitoring the spark activity, MRR and surface finish of the workpiece in the EDM process. Based on this, a reliable monitoring and diagnosis tool for EDM can be developed using the AE sensor and provides an opportunity to implement the industry 4.0 initiatives effectively. © 2018 The Author(s).