Chinese traditional musical instrument classification method based on depth confidence network

Abstract

The invention discloses a Chinese traditional musical instrument classification method based on a depth confidence network. During feature extraction of Chinese traditional musical instrument music, firstly, acoustic features of Chinese traditional musical instruments are extracted as primary features by employing a speech signal processing method, a depth learning network is constructed according to the depth confidence network, and more abstract features are extracted from the primary features of the Chinese traditional musical instruments by employing the depth learning network; and reconstruction abstract features of the Chinese traditional musical instruments are input to a softmax layer to predict the belonged type of the corresponding played musical instrument. The method is simple and feasible, the classification accuracy of the Chinese traditional musical instruments is improved, and more effective information is provided for the field of music information retrieval.

Description

technical field [0001] The invention belongs to the field of musical instrument classification and deep learning, in particular to a classification method of traditional Chinese musical instruments based on a deep belief network. Background technique [0002] With the development of computer networks and digital music, music data analysis and retrieval has become a hot research field in recent years. There are many content-based music information retrieval, such as music genre classification, singer identification, etc. Among them, the classification of musical instruments is also a very important field of music information retrieval, which is very important for the recognition and classification of music genres, emotions, scenes, etc. For example, when performing music genre classification, if the musical instrument used in the music file is known, using this information can improve the classification accuracy of the music genre. [0003] Although the classification of mus...

AI Technical Summary

Problems solved by technology

At present, most of the classification of musical instruments is based on manually selected acoustic features. Manually selecting features is time-consuming and laborious, and the features are diverse. It is difficult to find a unified model suitable for different objects for classification.

Images