I have a problem defining a deep model for extracting the raw signals from noisy signals. I have signals that create using fECG + mECG + noise as ECG signal and I also have fECG