Identifying archaeological ceramic pieces is a challenging problem for archaeologists, since fragments of archaeological pottery from the same site might have been made in different distant locations from the site. The pieces look very similar and contextual information is unavailable due to the small size of the fragments. Current standard methods to solve this problem are time consuming, require costly equipment, and can lead to the destruction of a part of the pieces. This work proposes a novel system that overcomes those limitations using non-destructive ultrasonic testing and advanced pattern recognition techniques. The proposed system was tested on real ceramic pieces from Aretina, Northern Italy and Sud-Gaul origin from archaeological sites of Paterna, and Les Jovaes in Valencia, Spain. The proposed fusion-based system achieved very high balanced classification accuracy and F1 score.