Application Of Non-Uniform Sampling In Compressed Sensing For Speech Signal

Currently, the most widely used Gaussian random observations in compressed sensing require that signals must be discrete, and the signal waveform must be known before observation, which greatly restricts the application of compressive sensing in speech. In response to this problem, this paper draws on the advantages of non-uniform sampling, constructs a non-uniform observation matrix, directly extracts the data from the signal waveform as observations, and gives a corresponding new method of reconstruction. The theoretical analysis and simulation results show that non-uniform observation can directly apply compressed sensing to analog speech signal processing, and the corresponding reconstruction method effectively enriches the means of compressive perception reconstruction.