A method and an apparatus for establishing a voice print model, a computer device, and a storage medium, the method comprising: combining voice acoustic features in a voice signal into a plurality of cluster structures; after calculating the average value and standard deviation of the plurality of cluster structures, performing coordinate transformation and an activation function calculation to obtain eigenvector parameters; and, on the basis of the eigenvector parameters, obtaining a voice print model. The present voice print model can reduce the voice recognition error rate of voice print models.