Multiple Musical Instrument Signal Recognition Based on Convolutional Neural Network

Abstract

To improve the accuracy of multi-instrument recognition, based on the basic principles and structure of CNN, a multipitch instrument recognition method based on the convolutional neural network (CNN) is proposed. First of all, the pitch feature detection technology and constant Q transform (CQT) are adopted to extract the signal characteristics of multiple instruments, which are used as the input of the CNN network. Moreover, in order to improve the accuracy of multi-instrument signal recognition, the benchmark recognition model and two-level recognition model are constructed. Finally, the above models are verified by experiments. The results show that the two-level classification model established in this article can accurately identify and classify various musical instruments, and the recognition accuracy is improved most obviously in xylophone. Compared with the benchmark model, the constructed two-level recognition has the highest accuracy and precision, which shows that this model has superior performance and can improve the accuracy of multi-instrument recognition.

Inferring Drumhead Damping and Tuning from Sound Using Finite Difference Time Domain (FDTD) Models

“…The fundamental time step is defined as ∆t, and the fundamental length of a cell is defined as ∆x = ∆y. To maintain stability, constants ∆t and ∆x are chosen such that no artificial energy is introduced (Courant-Friedrichs-Levy condition) [18,19] as shown in Equation (5).…”

“…The investigation of musical instrument timbre through machine learning methodologies is a well-established research approach. For example, musical instrument classification and recognition in single or multiple instrument performance recordings is a well-known algorithmic task in Music Information Retrieval [5]. However, the use of machine learning for the analysis of the vibroacoustic behavior of musical instruments is only found in more recent endeavors.…”