Vector Quantization Coefficient Indexing to Reduce Musical Noise

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing vector quantization methods, such as SVQ and AVQ, suffer from musical noise and block noise due to low amplitude quantization precision, which affects both frequency-domain and time-domain signals.

Innovation Solution

An encoding and decoding method that outputs index information indicating a group of coefficients minimizing the error between sample values and their quantized counterparts, allowing for reduced quantization error by adjusting decoded values with predetermined coefficients, thereby minimizing musical noise.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If spherical vector quantization (SVQ) method is used to quantize samples in units of sub-bands with dynamic bit assignment, then quantization precision for sparse signals is improved, but musical noise occurs when input signals have energy in many frequencies

Engineering Contradiction:
Improvequantization precisionVSAvoidmusical noise
Core Design Contradiction:
Measurement precisionVSObject-generated harmful factors

Solution Approach 1:

The patent divides the frequency spectrum into multiple sub-bands and processes each sub-band separately with dynamic bit assignment. This segmentation allows preferential quantization of dominant frequency components while using fewer bits for less important components, reducing overall quantization noise and preventing musical noise artifacts.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies different quantization precision to different sub-bands based on their perceptual importance and energy distribution. Sub-bands containing significant signal energy receive higher quantization precision, while sub-bands with minimal energy use lower precision, optimizing the balance between quality and bit rate while avoiding musical noise.

Inventive Principle:
Principle #3Local quality

2Loss of information

If algebraic vector quantization (AVQ) method is used to restore more frequency components, then completeness of decoded signals is improved, but amplitude quantization precision deteriorates

Engineering Contradiction:
Improvefrequency component restorationVSAvoidamplitude quantization precision
Core Design Contradiction:
Loss of informationVSMeasurement precision

Solution Approach 1:

The patent dynamically adjusts the number of bits allocated to each sub-band based on the signal characteristics and perceptual importance. This dynamic bit allocation allows the system to adaptively optimize between restoring frequency components and maintaining amplitude precision, depending on the specific signal conditions.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The patent changes the quantization parameters (number of bits per sub-band) based on the energy distribution and perceptual importance of different frequency components. By adjusting these parameters dynamically, the system can restore more frequency components when needed while maintaining adequate amplitude precision in critical sub-bands.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP2571170B1Encoding method, decoding method, encoding device, decoding device, program, and recording medium
Publication Date: 2016.06.01 NIPPON TELEGRAPH & TELEPHONE CORP
  • EP2571170B1 patent drawingFigure 1
  • EP2571170B1 patent drawingFigure 2
  • EP2571170B1 patent drawingFigure 3

AI summary

In encoding, index information indicating a group of coefficients that minimizes the sum of the error between the value of each sample and the value obtained by multiplying the quantized value of the sample by a coefficient corresponding to the position of the sample, for all sample positions, among a plurality of groups of predetermined coefficients corresponding to the positions of the samples, is output. In decoding, a plurality of values corresponding to an input vector quantization index are obtained as decoded values corresponding to a plurality of sample positions; and, with the use of a group of predetermined coefficients corresponding to the plurality of sample positions and indicated by input index information, the values obtained by multiplying the decoded values and the coefficients, corresponding to the sample positions are output.