Seismometers have precision on the order of ppb. Considering that there is no acoustic energy released prior to an earthquake, an infrasound precursor is extremely unlikely and something that seismologists would have already characterized.
This study says earthquake precursor movements send signals into the ionosphere, that cause other signals to propagate to earth, which can make objects vibrate audibly, minutes or possibly hours before the main shock. It argues this may be what animals are hearing. Rube Goldberg if true.
That would be my (completely unsubstantiated) guess as well. Animals, including humans, are surprisingly sensitive to infrasound, but there hasn't been a ton of research on the subject so far.