Abstract: Aiming at the sound transmission characteristics of the real human ear physiological structure are not considered in the loudness calculation of the existing vehicle interior noise sound quality prediction, a method for predicting the sound quality of vehicle interior noise based on the physiological loudness perception model is proposed. Firstly, the samples of vehicle interior noise from two cars were collected, and the subjective evaluation of annoyance of vehicle interior noise was obtained through subjective evaluation tests. Secondly, by combining the lumped parameter model of the middle ear and the cochlear transmission line model, a physiological loudness model was constructed. Thirdly, taking the calculated loudness of the physiological loudness model and the subjective evaluation values of vehicle interior noise as the main parameters, combined with sharpness, roughness, a sound quality prediction model was constructed by TabNet model. Finally, the prediction effects of the proposed sound quality model and the sound quality models based on the existing standard loudness models were compared. The results show that the average error percentage of sound quality prediction based on the physiological loudness model is only 4.73%, which is lower than that based on Moore loudness model (6.13%) and Zwicker loudness model (6.94%). Meanwhile, the average error percentage of the TabNet sound quality prediction model is also lower than that of the BP neural network prediction model (7.60%). The sound quality TabNet prediction based on the physiological loudness model can improve the accuracy of the objective evaluation of the sound quality of vehicle interior noise.