We developed an algorithm for high-quality, non-invasive maternal and fetal ECG (fECG) monitoring. We tested its ability to extract fECG from a single channel maternal thoracic ECG signal rather than the routine abdominal signal, and compare it with other algorithms. At 126dG, ECG was attached to near-term ewe and fetal shoulders, manubrium and xyphoid processes (n=12). Fetal ECG served as the ground-truth to which the fetal ECG signal extracted from the simultaneously-acquired maternal ECG was compared. The fECG extraction algorithm requires three steps. First, the de-shape short-time Fourier transform (STFT) is applied to estimate the maternal instantaneous heart rate, and hence maternal R peaks. Second, the nonlocal Euclidean median (NLEM) is applied to recover the maternal ECG. By direct subtraction, we obtain the rough fECG. Third, the fetal R peaks and fECG are obtained by applying the deshape STFT and the NLEM. All fetuses were in good health during surgery. In all animals, our algorithm and two widely applied single lead fetal ECG extraction algorithms failed to extract any fECG from the thorax maternal ECG signal with the F1 less than 50%. The applied fECG extraction algorithms might be unsuitable for the thoracic maternal ECG signal, or the latter does not contain strong enough fECG signal, although the lead is near the mother's abdomen. Fetal sheep model is widely used to mimic various fetal conditions, yet ECG recordings in a public data set form are not available to test the predictive ability of fECG and FHR. A challenge is the requirement to perform a sterile fetal surgical instrumentation with precordial ECG leads. We are making this data set openly available to other researchers and inviting others to share their maternal/fetal ECGs to foster non-invasive fECG acquisition in this animal model.