Advisors: Hongqing Liu [Undergraduate Graduation Design]
This project presents an overall design of a supervised deep learning based speech denoising scheme. An end-to-end speech denoising approach is adopted, and the data set is tested (noise reduction processing experiments) with the noise reduction scheme designed in this project. Through the testing of various objective indicators and the design of a series of subjective perception experiments, the effectiveness of the speech denoising model designed in this project is verified in an all-round way from two dimensions of objective quantitative analysis and subjective perception.
