Title: Research on singing breath correction system based on improved deep learning

Authors: Pengjiang Yu; Hui Tang

Addresses: College of Music, Chongqing Normal University, Chongqing 401331, China ' Art College of Hunan City University, Yiyang 413099, China

Abstract: The traditional system has some problems, such as low correction accuracy and long correction time. This paper designs a singing breath correction system based on improved deep learning. The rectification content is extracted by the singing audio extraction module, and the sample data is stored by the singing breath sample storage module. The improved deep learning algorithm is used to select a small part of the stored samples to input at the lowest level of the neural network. Combined with the probability density function, the distribution probability between the original audio signal and the input sample signal of the filter is determined, and the output of the audio signal after noise processing is processed to correct the singing breath. The experimental results show that the accuracy of the system is about 96%.

Keywords: improved deep learning; singing breath correction; filter; probability density function.

DOI: 10.1504/IJICT.2023.132763

International Journal of Information and Communication Technology, 2023 Vol.23 No.2, pp.164 - 176

Received: 10 Jun 2021
Accepted: 22 Jul 2021
Published online: 09 Aug 2023 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article