Frequency Shifting-based Variational Mode Decomposition Method for Speech Signal Decomposition

• Published in: 2022 International Conference on Automation, Robotics and Computer Engineering (ICARCE) pp. 1 - 5
• Main Authors: Liu, Wenyang, Hu, Weiping, Fu, Deli
• Format: Conference Proceeding
• Language: English
• Published: IEEE 16.12.2022
• Subjects: frequency shifting; mode mixing; speech decomposition; variational mode decomposition
• DOI: 10.1109/ICARCE55724.2022.10046652

In order to solve the problem of mode mixing and mode aliasing arising from speech decomposition, this paper proposes a speech signal decomposition method based on Variational Mode Decomposition (VMD): Variational Mode Decomposition-Frequency Shifting, VMD-FS). The method takes advantage of the VMD's good extraction of the fundamental frequency of the speech signal, sets specific carrier parameters to shift the frequency of the speech signal to lower frequency, and then applies specific parameters and iterative methods to the VMD to decompose the speech signal in order to obtain the true IMFs that make up the speech signal. Through the decomposition experiments of real speech signals, it is demonstrated that VMD-FS solves the phenomenon of mode mixing and mode aliasing issues arising from the decomposition of speech signals compared with Empirical Mode Decomposition (EMD) and the original VMD method. From the Mean Square Error (MSE) of the decomposition results of the above three methods, it can be proved that VMD-FS outperforms EMD and VMD methods