Prosthetic limbs (and orthotic devices) have been used as a paradigm for the treatment and rehabilitation of neuropathic pain, such as phantom limb pain. Long-term adoption of the devices for the continued use in rehabilitation remains low in part due to reduced embodiment and the high cognitive load associated with controlling the device. Previous research has shown that incorporating sensory feedback in prostheses can provide proprioceptive information, increase control and manipulation of objects, and improve embodiment. However, feedback experienced by the user varies daily and requires constant parameter adjustments to maintain accurate and intuitive sensory perception, further preventing long term adoption. Work therefore needs to be explored that correlate feedback modalities to perception of tactile information, such as texture and pressure. The work presented in this paper begins to explore this by utilizing a deep-learning algorithm to classify the dissipation of vibration artefacts found in the EMG signals of able-bodied individuals to specific texture patterns. Four texture patterns were applied to 7 participants using two vibration motors and repeated 3 times. In post processing, a RNN network identified the artefact features along equidistantly spaced EMG electrodes and correctly classified unseen data from each participant.