SNN (Spiking Neural Networks) has long been a popular research area in deep learning, combining neuroscience and machine learning to create models that most closely simulate the mechanisms of biological neurons for computation. However, the model is still in the exploratory research phase. This paper explores the use of SNN models to complete a real-world application—Real-time Speech Recognition using Spiking Neural Networks. The STM32 MEMS microphone is used as the sound input in this study, and MFCC is applied to process the audio data, which is then converted into spike encoding for SNN training. The trained model is deployed on the PYNQ board for real-time speech recognition.