Compressed sensing reconstruction method for speech signals
A technique of compressive sensing reconstruction and voice signal, which is applied in voice analysis, instruments, etc., can solve the problems of slow calculation speed, difficult to satisfy, and high computational complexity, and achieve the goal of less reconstruction time, high matching degree, and small calculation amount Effect
Active Publication Date: 2017-03-29
NANJING UNIV OF POSTS & TELECOMM
View PDF6 Cites 6 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
The above two algorithms have the following characteristics. The greedy algorithm is simple to calculate and has a good reconstruction effect. However, most greedy algorithms require that the sparsity of the signal is known, which is difficult to satisfy in practice.
Convex optimization algorithm has a high signal reconstruction rate and requires few observation points, but has high computational complexity and slow calculation speed
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0024] The present invention provides a method for compressed sensing reconstruction of speech signals, which now combines figure 1 The given general process of compressed sensing and the intermediate results of the experiment will discuss in detail the implementation process and innovations of the present invention:
[0025] Step 1: Obtain the observed signal and solve the sparse solution
[0026] 1. The present invention is a male voice "LSI" of 1s selected from the Chinese voice bank of the Institute of Automation, Chinese Academy of Sciences. The sampling frequency of this signal is 16kHz, and each frame signal has 256 sampling points;
[0027] 2. Select the Gaussian observation matrix for observation. The form of the matrix is as follows:
[0028] Construct a matrix Ψ with a size of M×N, where M is the dimension of the observed signal, N is the dimension of the original signal, and each element in Ψ independently obeys a Gaussian with a mean value of 0 and a variance o...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention discloses a compressed sensing reconstruction method for speech signals. The method overcomes the deficiency of the existing speech signal compressed reconstruction technology. A speech signal compressed reconstruction method based on a Smooth L0 norm is proposed. Compared with the traditional speech signal compressed reconstruction method, the SL0 algorithm used in the method of the invention does not need to know the sparse degree of a speech signal before reconstruction, and has the advantages of small amount of calculation, high degree of matching, less reconstruction time, and the like. In order to achieve the purpose, a new Smooth L0 norm is used to reconstruct a speech signal. Compared with the traditional speech signal reconstruction method, the proposed improved smooth L0 algorithm uses a steepest descent method and a gradient projection algorithm, and thus has the advantages of small amount of calculation, high degree of matching, less reconstruction time, and the like.
Description
technical field [0001] The invention relates to a compression sensing reconstruction method of a voice signal, belonging to the technical field of voice signal compression sensing. Background technique [0002] The processing of speech signals in traditional methods is based on the Nyquist sampling theorem. The theorem points out that the sampling frequency must be greater than twice the highest frequency of the signal, otherwise the original signal will not be well restored at the receiving end. The theorem gives The relationship between the sampling frequency and the spectral distribution of the signal is a sufficient, but not necessarily necessary, condition for accurate reconstruction of any signal. How to remodel the speech signal according to the particularity of the speech signal to obtain fewer samples without affecting the reconstruction quality of the speech is a research hotspot in the field of speech signal processing. [0003] The compressed sensing theory prop...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More IPC IPC(8): G10L19/00
Inventor 孙林慧赵城薛海双
Owner NANJING UNIV OF POSTS & TELECOMM