Since their inception, the objective of digital communication systems has been to effectively and reliably transmit a series of bits representing a certain type of information. Consequently, all engineering choices made in past years to construct such systems pursued this goal, overlooking the semantic significance of the information intended for transmission. Semantic communications, viewed as a potential breakthrough in the Shannon paradigm, aim to convey the meaning of a message rather than accurately transmitting every symbol. Recent advancements in machine learning have enabled the extraction of such information from signals, a development that has the potential to revolutionize signal processing techniques in the future.