Quantizer code rate distortion controlling means based on advanced audio coder

Abstract

The invention discloses a quantizer code rate distortion control method based on an advanced audio encoder. The method includes the following steps: A, obtaining the pre-estimated number of the distributable bits of the sub-band through the code stream to be encoded, and calculating the average number of the sampling distributable bits according to the pre-estimated number of the distributable bits of the sub-band; B, determining the selecting coefficient of the rift belt quantization scale based on prearranged principle through the average number of the sampling distributable bits, and obtaining the rift belt quantization scale through calculation through the improved discrete cosine transform spectrum of the code stream to be encoded and the selecting coefficient of the rift belt quantization scale; C, conducting code rate distortion control with the rift belt quantization scale as the parameter and obtaining the final reference bit consumption. The quantizer code rate distortion control method effectively reduces the complexity of code rate distortion control and reduces the calculation quantity during quantizing coding, thus improving encoding quality and efficiency of the encoder.

Description

A Quantizer Rate Distortion Control Method Based on Advanced Audio Encoder technical field The present invention relates to a quantizer code rate distortion control method based on Advanced Audio Coding (AAC). Background technique AAC is a type of transform domain coding whose goal is to optimize a certain measure of distortion under certain bit constraints. Distortion is caused by the quantization process. During the quantization process, the rate-distortion control is assigned to the parameters of the quantization encoder to obtain the exact number of bits consumed, so that the quantization distortion can be controlled within the perceptual tolerance range, so as to ensure the coding accuracy. quality and efficiency. AAC coding uses non-uniform quantizers and variable-length coding, which makes the design of quantization algorithms very difficult. First, a non-uniform quantizer cannot formulate encoder parameters based on perceptually tolerant noise, and, due to varia...

AI Technical Summary

Problems solved by technology

The double-loop nested search (TLS) algorithm is a widely used R-D control algorithm at present, and it is the algorithm recommended by the AAC standard, but this algorithm will bring the unrealization of the algorithm when it is forced to search for the optimal solution. The complexity of O((60′11)49) (60 quantization order possibilities, 11 codebooks, 49 independent subbands)

In order to reduce the complexity of the R-D algorithm, several correction methods have been proposed, including setting the maximum number of cycles, narrowing the search range of the quantization order, using a statistical linear model to predict bit consumption, using the parameters of the previous frame as the initial value of the current frame, etc. , but did not achieve good results

Images