Successful deep learning-based approaches for tool wear condition monitoring (TWCM) need large sample labelling and a consistent probability distribution across training and testing data. But these two requirements are not often easy to meet in real experimental data; thus, those approaches’ performance takes a major hit. This paper proposes generation of training and testing data for training of artificial neural network (ANN) using response surface model (RSM). Sound emission and tool vibrations signals based on multisensory approach are utilized along with milling parameters for developing RSM model. By utilizing analysis of variance (ANOVA), we were able to statistically validate the response model and determine the most efficient inputs for a set of milling parameters. The model’s efficacy within the specified domain of machining parameters is demonstrated by the correlation coefficient (R2) value of 99.2%. Furthermore, data are generated for training and testing using this validated RSM. ANN model is evaluated by mean square error (MSE) which is 4.2979 e-6 and regression R value which is 0.99. The model prediction in testing has shown good accuracy which reveals that considering the sound pressure of machining and tool vibration as input parameters for the model development can help improve the accuracy of tool wear prediction.