# US 8,149,927 B2 · Method of and apparatus for encoding/decoding digital signal using linear quantization by sections

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 8,149,927 B2
- Document ID: US8149927B2
- Status: Granted
- Kind: B2
- Application number: 12/792,048
- Filed: 2010-06-02
- Published: 2010-09-23
- Granted: 2012-04-03
- Earliest claimed: 2004-05-12
- Assignees: Samsung Electronics Co., Ltd.
- Inventors: Junghoe Kim; Dohyung Kim; Shihwa Lee; Sangwook Kim
- Web page: https://patentgrep.com/patent/US8149927B2

## CPC

- G10L 19/032: Quantisation or dequantisation of spectral components
- G10L: SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING

## Abstract

A method of encoding/decoding a digital signal using linear quantization by sections, and an apparatus for the same are provided. The method of encoding includes: converting a digital input signal, and removing redundant information from the digital signal; allocating a number of bits allocated to each predetermined quantized unit considering the importance of the digital signal; dividing the distribution of signal values into predetermined sections based on the predetermined quantized units, and linear quantizing data converted pin the operation of converting the digital input signal by sections; and generating a bit stream from the linear quantized data and predetermined side information. Therefore, a sound quality is improved compared to a sound quality produced by conventional linear quantizing devices and a complexity of a non-linear quantizing device is reduced.

## Claims

### Claim 1

A method of encoding a digital signal, including predetermined side information, using linear quantization by sections, the method comprising:
removing, performed by at least one processing device, redundant information from the digital signal by converting the digital signal;
allocating a number of bits to each predetermined quantized unit of the digital signal in consideration of an importance of the digital signal;
dividing a distribution of the digital signal values into predetermined sections;
linear quantizing data that is divided into the predetermined sections by using a linear function set for each of the sections, the sections divided in consideration of the distribution of digital signal values; and
generating a bit stream from the linear quantized data and the predetermined side information.

### Claim 2 (depends on 1)

The method of claim 1, wherein the dividing of the distribution of signal values and linear quantizing of the data comprises:
normalizing data converted in the converting of the digital signal using a predetermined scale factor for each of quantized units;
dividing a range of normalized values into predetermined sections, and converting the normalized data using a linear function that is set for each of the predetermined sections;
scaling a value converted in the operation of converting the normalized data using the number of bits allocated to each quantized unit; and
calculating a quantized value by rounding the scaled value in the scaling operation.

### Claim 3 (depends on 2)

The method according to claim 2, wherein, if the normalized data is not uniformly distributed, a range of the normalized data is divided into predetermined sections.

### Claim 4 (depends on 2)

The method of claim 2, wherein the scaling factor is an integer determined by a predetermined function of a value that is greater than or equal to an absolute maximum value after calculating the absolute maximum value among sample data values within the quantized unit.

### Claim 5 (depends on 2)

The method of claim 2, wherein the linear function used in the dividing of the range of normalized values is expressed as a plurality of independent linear functions for each predetermined section.

### Claim 6 (depends on 5)

The method of claim 5, wherein the dividing of the range of normalized values and the converting of the normalized data comprises:
dividing the range of normalized values into two sections; and
converting the normalized data by applying a linear function set for each of the sections to the data,
wherein the linear functions are
y=αx/(α−2b
and
y=x/(1+2b)+2b/(1+2b)
(here, a denotes the range of normalized values, and b denotes a section displacement from the center of a).

### Claim 7 (depends on 2)

The method of claim 2, wherein the linear function is continuous.

### Claim 8 (depends on 1)

The method of claim 1, wherein the converting of the analog signal is performed by one of a discrete cosine transform, a fast Fourier transform, a modified discrete cosine transform, and a subband filter.

### Claim 9 (depends on 1)

A computer readable recording medium storing a program to execute the method disclosed in claim 1.

### Claim 10 (depends on 1)

The method according to claim 1, wherein the input data is a pulse control modulation (PCM) audio signal.

### Claim 11 (depends on 1)

The method according to claim 1, wherein the removal of the redundant information is performed by subband filtering.

### Claim 12 (depends on 1)

The method according to claim 1, wherein the importance of the digital signal is based on a psychoacoustic perspective relating to hearing characteristics of humans such that more bits are allocated to frequencies to which humans are highly sensitive.

### Claim 13 (depends on 1)

The method according to claim 1, wherein in the allocating of a number of bits to each predetermined quantized unit of the digital signal, detailed information with low sensitivity is omitted using hearing characteristics of humans, a bit allocation amount is set differently for each frequency so as to reduce the encoding amount, and bit allocation information is calculated considering a psychoacoustic perspective.

### Claim 14

An apparatus to encode a digital signal, including predetermined side information, using linear quantization by sections, the apparatus comprising:
a data converting unit to remove redundant information from the digital signal by converting the digital signal;
a bit allocating unit to calculate a number of bits to be allocated to each predetermined quantizing unit of the digital signal considering an importance of the digital signal;
a linear quantizing unit to divide a distribution of data values into predetermined sections and to linear quantize data that is converted by the data converting unit by using linear function set for each of the sections, the sections divided in consideration of the distribution of digital signal values; and
a bit packing unit to generate a bit stream including the linear quantized data generated by the linear quantizing unit and the predetermined side information.

### Claim 15 (depends on 14)

The apparatus of claim 14, wherein the linear quantizing unit comprises:
a data normalizing unit to normalize the data converted by the data converting unit using a predetermined scaling factor;
a section quantizing unit to divide a range of normalized values into predetermined sections, and to apply a linear quantizing function set for each of the sections to the normalized data;
a scaling unit to scale values generated by the section quantizing unit using the number of bits allocated by the bit allocation unit; and
a rounding unit to generate a quantized value by rounding the scaled value based on the number of allocated bits.

### Claim 16 (depends on 14)

The method of claim 14, wherein the converting of the digital signal is performed by one of an inverse discrete cosine transform, a fast Fourier transform, a modified discrete cosine transform, and a subband filter.

### Claim 17 (depends on 14)

The apparatus according to claim 14, wherein the input data is a pulse control modulation (PCM) audio signal.

### Claim 18 (depends on 14)

The apparatus according to claim 14, wherein the removal of the redundant information is performed by subband filtering.

### Claim 19 (depends on 14)

The apparatus according to claim 14, wherein the importance of the digital signal is based on a psychoacoustic perspective relating to hearing characteristics of humans such that more bits are allocated to frequencies to which humans are highly sensitive.