Cardiotocography provides information on uterine shrinkages with fetal heart rate, that is highly helpful in determining if fetus is normal, questionable, or pathologic, it is used to measure FHR in fetus throughout pregnancy to assure physical health. Due to human error, several cardiotocography measures make incorrect inferences and predictions. It is crucial to monitor the fetus’s status at various stages and to provide it with the right medication for its well-being. In the current work, Recurrent Neural Network model and frequency time representation of FHR signal is used to generalize Morse wavelet which constructs a unique computer-aided fetal distress diagnostic method. Following preprocessing, the FHR signal’s time frequency information is extracted by Morse wavelet and put forth to a ResNet 50 model that has already been trained and fine-tuned in accordance with the dataset. The model taken from the binary confusion matrix is measured for sensitivity, specificity, and accuracy. A range of signal processing methods are employed to adjust the retrieved CTG signal. The CTG signal is then broken down into its frequency components using empirical mode decomposition (EMD), and features related to instantaneous frequency and spectral entropy are then retrieved. Moreover, feature selection and normalization using the Relief method to classify data into normal and pathological categories based on the experimental results proposed method achieved an AUC of 91% compared to state of art methods on various CTG datasets.