Spiking neural networks only pass binary signals between neurons, unlike classical neural networks that pass a scalar activation. This is closer to biological neurons and may enable advantages of the brain over machines like energy efficiency and learning from a few examples. Neuromorphic hardware is at the frontier of achieving these benefits in practice; however, many machine-learning tasks are yet to be implemented this way. This research aims to develop a speech recognition neural network on Loihi 2 neurocore and test its accuracy and efficiency against classical approaches in noisy environments.