Plant stress is a physiological response triggered by various environmental factors, such as extreme weather conditions, diseases, or mechanical disturbances. Monitoring and classifying plant stress are crucial for assessing the overall health of vegetation and implementing timely interventions to ensure sustainable agricultural practices. However, estimating the magnitude of stress proves challenging due to a lack of comprehensive datasets and limited technological capabilities. Therefore, in this study, we present a plant classification method leveraging a 2D representation based on Short-Time Fourier Transform of 6-channel 500 kHz sounds, recorded in scenarios such as tree cutting, drought, and other common environmental conditions. The experimental results demonstrate that our proposed method effectively predicts stress factors; however, due to a lack of a comprehensive dataset and limitations in data quality, the accuracy is slightly lower. These endeavors are expected to offer significant assistance in the analysis and judgment of plant stress.