Piano teaching should not continue to be limited to the mere cultivation of students’ performance skills but should be integrated into curriculum thinking and politics while following the characteristics of the profession. This paper conducts piano teaching experiments through deep learning LSTM algorithm and extracts a series of Civic and Political elements from them as a fitting point of piano teaching and curriculum Civic and Political to discuss, using multiple teaching modes to strengthen the cultural transmission of piano courses to improve students’ ideological and moral level and comprehensive quality. The experimental results show that the accuracy of the deep learning LSTM algorithm based on the integration of the deep learning piano course curriculum Civics is stable up to 98.17%, the average false detection rate is stable at 1.97%, the average positive recall rate is stable at 94.7%, and the negative recall rate is 73.2%, and the LSTM algorithm improves the integration degree by 31% compared with the traditional SVMSGD and RF algorithms. It shows that carrying out experiments of deep learning based on deep learning for teaching Civics in college piano courses effectively strengthens the cultural transmission of piano courses and improves students’ ideological and moral level and comprehensive quality.