The goal of this study is to provide objective analysis and quantitative research for the auscultation in Traditional Chinese Medicine (TCM) by taking advantage of algorithm of wavelet packet and Deep Forest. Based on wavelet packet transform, the speech signals are decomposed into six-layer wavelet packet coefficients. Shannon entropy values are extracted as a useful acoustic parameter from wavelet packet coefficients, which are employed for feature analysis and recognition of healthy, Qi-deficiency and Yin-deficiency subjects. After the feature vector formed by the Shannon entropy values are inputted into Deep Forest to be trained and predicted. The results showed the methods proposed were effective and efficient to analyze and recognize auscultation signals. Although the number of subjects is limited, the classified results are satisfied in summary