Firmness can be used as one of the important indexes to indicate the ripeness of apple, usually determined by acoustic signal detection nondestructively, for the high correlation between acoustic signal features and firmness. In this paper, acoustic signal features with high time correlation are selected as the independent variable of the model. The acoustic signal and quality of apple were measured, and then 32 time-dependent signal features were selected by MATLAB analysis data, and the firmness of apple was predicted by ridge regression model. Using this method, the accuracy of training set and verification set is 84.7and 82.4% respectively. The actual change of apple is consistent with the prediction trend. Because of its more characteristics and more accurate model, this method has better fault tolerance for data, reduces the environmental requirements for acoustic signal detection, and paves the way for the firmness prediction of other apple varieties.