Speech Command Recognition Algorithm Based on Improved MFCC Features

In recent years, deploying low-power speech recognition algorithms on resource-constrained hardware devices has become a research focus. However, traditional algorithms rely on extensive data and complex models, which pose challenges for low-power devices with limited resources. This paper proposes an improved MFCC feature extraction method combined with the DTW algorithm, which enhances computational speed and reduces resource consumption in speech recognition. Experimental results validate the effectiveness and feasibility of the proposed method.