Milling-Electrochemical Spark Micromachining (M-ECSMM) is a cutting-edge hybrid micromachining technique that produces channels, grooves, and features. The experiment performed in the current work uses the one-factor-at-a-time method (OFAT) to build microchannels on glass and silicon substrates. Then the data obtained is trained using an artificial neural network (ANN). Developed NN model predicts MRR and Ra using input parameters such as voltage, tool speed, electrolyte concentration, and pulse on-time (Ton), keeping pulse off-time (Toff) constant. The analysis of a comparative parametric plot of predicted values MRR and Ra for silicon and glass is done, which shows that predicted values of MRR for silicon maintain the same level with a small rise in values. Still, for glass, it shows a significant rise at 12 rpm, 95 V, Ton as 900 µs, and 4.6 M concentration. For the glass workpiece, the trained model under the same input process parameters gives MSE for training as 2.118, which is greater than the silicon workpiece, while for testing and validation, it is 8.43 and 4.00, respectively. Also, it shows the overall MSE and regression values of the NN model developed for both workpieces, and from the results, we observe that error is more in the case of glass than silicon workpieces. Thus, we can say that the model presented in this paper for process parameters best fits a silicon workpiece compared to a glass workpiece.