The heart sound split recognition method based on multiple heart sounds comprises the following steps: synchronously collecting and synchronously recording electrocardiograms and multi-channel phonocardiograms, respectively identifying the electrocardiographic characteristics of each cardiac cycle; correspondingly finding each path of the phonocardiogram according to each electrocardiographic feature Heart sound characteristics, respectively for each cardiac cycle in the electrocardiogram to obtain the R wave, the S1 corresponding to the current R wave on the M area phonocardiogram, the time corresponding to M1, and the T area on the T area phonocardiogram. S1 Marked S1, obtain the time corresponding to T1, and in each phonocardiogram, obtain the interval time T between M1 and T1 for each cardiac cycle M‑T , if T M‑T =0.02S, then S1 is a normal natural split; when a T appears in a phonocardiogram M‑T >0.02S, then S1 wide division, abnormal. The invention can automatically identify the natural split or abnormal split of the heart sound, and is suitable for home monitoring.