Speech separation, also known as source separation or blind source separation, refers to the process of separating individual speech signals from an audio mixture that contains multiple overlapping speakers. This technology has numerous applications across various domains, e.g., speech enhancement, automatic speech recognition, speaker diarization, hearing aids, forensic analysis, and music remixing and production. In this study, we build different deep-learning models to perform on the sound separation problem. We have shown that the deep learning method is significantly effective and superior to traditional methods. Besides, we also compare the effectiveness of different deep learning methods on single-channel speech separation