Recording of acoustic emissions (elastic waves emitted by dislocation motion) during compression of microscopic samples is a novel emerging tool for the investigation of mechanical properties of materials. Yet the resulting signals can be so noisy and complex that they don't easily allow for identifying the deformation mode or the material involved. This study aims at testing if the human auditory system can pick patterns in the audified recordings (i.e. transposed from the ultrasonic to the audible range) that match the physical characteristics of the samples. In a Free Sorting Task, participants sorted sounds into groups according to perceived similarity. The sounds originated in recordings varying according to material, deformation mode, and size of the sample. During clustering and multidimensional analysis, sounds were found to be first grouped according to spectral features: Zinc and Magnesium recordings respectively produced high-pitched and low-pitched sounds. Sounds were further discriminated according to perceived sound level (soft vs. loud for slipping vs. twinning) and number of perceived sound events (no clearly associated physical parameter). Correlations were found between subjective groupings and audio descriptors, so that audio-inspired mechanical investigations can now be envisioned.