In this work, our contribution will intervene to reduce the impact of noises on the ECG signals. Various ECG denoising approaches were tested to see how efficient they were in removing dominant noises that add to pure ECG signals. Due to different causes such as interference, muscular noise, body movement related to breathing, and so on, the original signal acquired by the electrodes produces noises. In this article, the electrode signals are monitored using an Internet of Things system that combines an Arduino board and an AD8232 module to generate a one-dimensional signal. These ECG signals are displayed on a computer using the Matlab interface. Following that, an efficient deep learning model was developed to facilitate cardiologists in their diagnosis of ECG signals. These experimental results obtained demonstrate the effectiveness of our proposed model compared to other existing methods in the literature. Finally, the filtered and classified ECG signals are given to the doctor for correct treatment of the patient's condition.