Unlock instant, AI-driven research and patent intelligence for your innovation.

Method and apparatus for multi-stage quantization

A quantization error and vector technology, applied to electrical components, speech analysis, instruments, etc., can solve the problem of waste of storage space, achieve the effect of saving storage space and improving resource utilization

Active Publication Date: 2008-12-31
HUAWEI TECH CO LTD
View PDF0 Cites 10 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0008] The problem to be solved by the embodiments of the present invention is to provide a multi-level quantization method and device to overcome the defect of wasting storage space in the prior art due to the independence of the codebooks of each level of quantization

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 more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Method and apparatus for multi-stage quantization
  • Method and apparatus for multi-stage quantization
  • Method and apparatus for multi-stage quantization

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0019] Below in conjunction with the accompanying drawings and examples, the specific embodiments of the present invention will be described in further detail:

[0020] In speech coding based on linear prediction, it is necessary to vector quantize the linear prediction parameters, taking the conversion of the linear prediction parameters into 10-dimensional LSF (Linear Spectrum Frequency, line spectrum frequency) parameters to perform two-level quantization as an example.

[0021] A flowchart of a two-level quantization method according to an embodiment of the present invention is as follows: figure 2 shown, including the following steps:

[0022] The scaling factor (set) γ needs to be obtained in advance and can be obtained by the following training method:

[0023] γ needs to be obtained by training with a large number of samples, and the simplest training method is obtained by comparing the space radius of the quantization error vector after the first-stage quantization ...

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

No PUM Login to View More

Abstract

The invention discloses a method for multistage quantification which comprises the steps as follows: a reference codebook is obtained according to a pre-stage codebook; a current-stage codebook is obtained according to the reference codebook and scaling factors; input vectors are quantized by utilizing the current-stage codebook. The invention also discloses a device for multistage quantification. Through the relation between the current-stage codebook and the pre-stage codebook, the invention can obtain the current-stage codebook according to the pre-stage codebook, thus leading the current-stage codebook to require no corresponding independent codebook space, saving storage space and improving resource utilization rate.

Description

technical field [0001] The present invention relates to the technical field of data compression, in particular to a method and device for multi-level quantization. Background technique [0002] With the rapid development of data compression technology, vector quantization has been widely used. Vector quantization is an efficient data compression technique. This technique forms a vector with several scalar data columns, and then quantizes it as a whole in the vector space, thereby compressing the data without losing much information. The process is as follows: each frame of the k sampling points of the signal waveform or each parameter set with k parameters is formed into a vector in the k-dimensional Euclidean space, and then the vector is "collectively" quantized. In vector quantization, the k-dimensional infinite space is divided into M region boundaries, and then the vector of the input signal is compared with these boundaries and quantized as the central vector value of...

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
no application Login to View More
IPC IPC(8): G10L19/14G10L19/12G10L19/06G10L19/07
CPCG10L19/06G10L19/07G10L2019/0005G10L2019/0007H03M7/3082G10L19/12
Inventor 艾雅·舒默特代金良殷福亮马欣张军
Owner HUAWEI TECH CO LTD