Influenza viruses cause illness, and influenza viruses are especially dangerous. These viruses are notorious for spreading quickly, having the capacity to harm people seriously, and frequently changing, all of which highlight the necessity of further study and monitoring. A virus’s ability to spread between species is critical for predicting epidemics and creating potent vaccines. Historically, influenza pandemics between species have been brought on by viruses. To stop an outbreak from spreading, it is critical to determine the virus’s origin. Machine learning techniques to generate quick and precise predictions for viral sequences have recently gained popularity. Hemagglutinin sequences were the only ones used because they are the main protein implicated in the immune response. They were portrayed using the word embedding and a position-specific scoring matrix. The findings demonstrate that the convolutional neural network is the most effective algorithm for predicting viral sequence origins, achieving about 98.54% area under the precision-recall curve (AUCPR), 97.01% F1 score, and 95.60% Matthews’ correlation coefficient (MCC) at a higher classification level, and approximately 95.74% AUCPR, 89.41% F1 score, and 82.79% MCC at a lower classification level.