Design And Implementation Of Music Learning System Based On Comparison Of Characteristics

Music learning has a negligible effect on the future development of students. However, the current status of music teaching does not meet our expectations. Due to the special nature of the discipline of music, class teaching system as a kind of one-to-many teaching model, the disadvantage of which is even more serious. Especially in the stage of basic education, students’ music level is generally not high. And each student has different perception, and different learning progress. If teachers conduct teaching from the average level of students, it will be impossible to meet the learning needs of all students. In addition, in some remote areas which lack of qualified teachers, the number of music courses is very limited, students’music literacy can’t be promoted effectively. In this situation, we are badly in need of a new way to help even replace music teachers in singing teaching.The rapid development of digital audio technology provides a new solution to enhance the students’music literacy and singing skills. Through the recording, analysis and processing to the singing of students, we can evaluate automatically and objectively according to the professional evaluation standard, which can help the student practice singing more efficiently. For this purpose, we have developed a set of musicology sing system based on feature matching, which can be achieved for grading automatically through melody feature extraction, similarity matching and pitch data analysis, pointing out shortcomings for the students, analyzing the cause of the error, and providing some improvements suggestions.The core of this study lies in the design and implementation of the feedback on learning. Around this core, I did the following research and exploration:(1) The realization of Web real-time recording. The Web real-time recording has been realized through a combination of Flash and FMS server, and the recorded audio can be saved as WAV format.(2) The Research of pitch extraction. In this paper, both the singing voice and MIDI music in the template library have been extracted the pitch in different ways, generated the pitch characteristic sequence, so that we can compare its similarity.(3) The presentation of feedback information from singing/playing..In order to make students more intuitive to recognize and correct the mistake in singing, we have designed a feedback mode which takes the pitch and rhythm as the judging basis, and combines image, sound and text.In this paper, there are some innovative points:(1) On the function design: First, to make up the lack of educational factors in other existing similar systems, we pay more attention to whether the feedback of our system is diversified, detailed, effective enough or not, only by this way we can help students to enhance their singing level. Then, we add music resource categories. In addition to the music with human voice, there is a lot of pure music in the music library of system which can be used by students who need to take practice instrumental performance.(2) On the technological realization: First, taking three indicators as the judgment standard based on pitch sequence data, which is pitch, rhythm and breathe smoothness. Then, a rule-based smoothing method is put forward, which can get a better smoothing effect combining with the algorithm of the median five-point smoothing. Last, in the respect of similarity comparison, pitch data should be structured before cosine similarity algorithm, so that the result is more in line with the actual situation.