GB/T 33475.3-2018Information technology—High efficiency media coding—Part 3: Audio (English PDF)

3.1 Reserved

Temporarily unused fields in the defined encoded bitstream may be used in future standard extensions.

3.2 Bit rate

The rate at which the encoded bit stream is transmitted to the input of the decoder.

3.3 Bit stream

A set of bits that are used in a certain order as a data encoded representation.

3.4 Coding

The audio sample stream is read in and a valid bit stream conforming to this section is generated.

3.5 Encoder

Encoding processed entity.

3.6 Code representation

A unit of data represented in its encoded form.

3.7 Coded audio bit stream

An encoded representation of the audio signal.

3.8 Side information

The necessary information to control decoding in the bitstream.

3.9 Sampling frequency; fs

The number of samples of the discrete signal extracted from the continuous signal per second, which can be referred to as the sampling rate.

Note: The unit is Hertz (Hz).

3.10 Auxiliary data

Data used in the bitstream to assist in channel coding.

3.11 Decoding

One type of data processing defined in this section is the process of reading in an encoded bit stream and outputting audio sample values.

3.12 Decoder

Decode the processed entity.

3.13 Spectral coefficient

Analyze the discrete spectral domain data output in the filter bank.

3.14 Entropy coding

A variable length lossless coding in the digital representation of the signal to reduce redundancy in statistical properties.

3.15 Channel

Independent audio signals that are captured or played back in different spatial locations during recording or playback.

3.16 Data unit

A representation of the pre-encoded and encoded data items.

3.17 Fill (bit); fill (byte)

A codeword that can be inserted into a special location of the encoded bitstream and removed during the decoding process. In addition, the encoded additional data can also Use padding bits or bytes.

3.18 Signal type

A mechanism for classifying encoded audio signals for selecting different filtering methods and encoding methods.

3.19 Audio buffer

A buffer for storing encoded audio data in the decoder.

3.20 Byte

A sequence of 8 bits.

3.21 Byte aligned

In the encoded bit stream, the number of bits is a multiple of eight.

3.22 Noise coding

The statistical characteristics of the signal are similar to the degree of noise parameter coding.

3.23 Adding individual line

Add a sinusoidal component to a particular frequency band.

3.24 Linear predictive coding

An algorithm for processing an input audio signal to reduce signal redundancy and improve coding efficiency.

3.25 Lifting wavelet

Wavelet transform implemented with a lifting strategy.

3.26 Multichannel decorrelation

Inter-channel correlation is removed to improve coding efficiency.

3.27 Channel coding

The encoding of the base channel, ie the encoding of the underlying sound signal in addition to the sound object.

3.28 Sound object

A sound that is perceived as a whole or an environment-independent sound emitted by a sound source.

3.29 Sound object encoding

The sound object audio sample stream and its side information are read in and an encoded bit stream including object metadata and audio content is generated.

4.1 Arithmetic Operators

The following arithmetic operators are available for this document. plus - minus (binary operator) or negation (unary prefix operator)

× multiply The ab power, which represents the b-th power of a, can also represent the superscript. ^ power % modulus operator, defined only for positive integers.

/ Integer division, the result is rounded to zero. For example, 7/4 and -7/-4 are rounded to 1, 7/4 and 7/-4 rounded to -1. Division, no rounding or rounding.

|| Absolute value |x|=x when x >0 |x|=0 when x=0 |x|=-x when x< 0 Abs absolute value Sign() takes the symbol sign(x)=1 when x >0 Sign(x)=0 when x=0 Sign(x)=-1 when x< 0 x square root f(i) The cumulative sum of the functions f(i) when the argument i takes all integer values from a to b (including b).

Log10 base 10 logarithm Log2 base 2 logarithm Round down Exp an exponential function based on the natural constant e Residual residual, the difference between the actual observed value and the estimated value (fitted value)

4.2 Logical Operators

The following logical operators are suitable for this document. || Logical or Logical non

4.3 Relational operators

The following relational operators apply to this document.

> greater than >= greater than or equal to < less than <= less than or equal to != not equal Max[,,] maximum value in the parameter table Min[,,] the minimum value in the parameter table

4.4 Bit operator

The following bit operators apply to this document.

| or ~ Invert a > >b Shift a to b to the right in the form of a two's complement integer. This operation is defined only when b takes a positive number. To the right When moving to the most significant bit, its value is equal to the most significant bit before the a shift operation.

When moved to the least significant bit, its value is equal to zero.

4.5 Assignment

The following assignment operations apply to this document.

== assignment operator x = ab x takes the value from a to b (inclusive b), where x, a and b are integers.

Self-added, x is equivalent to x=x 1. When used for array subscripting, the variable value is evaluated before the self-addition operation.

-- Self-decreasing, x--equivalent to x=x-1. When used for array subscripting, the variable value is evaluated before the decrement operation.

= Self-added specified value, for example, x = 3 is equivalent to x = x 3, and x = (-3) is equivalent to x = x (-3).

-= Decrement the specified value, for example, x-=3 is equivalent to x=x (-3), and x-=(-3) is equivalent to x=x-(-3).

4.6 Mnemonic

The following mnemonics apply to this document. Rpchof polynomial remainder, high order first.

Bslbf bit string, the left bit is first, here "left" is the order of the bit string written in GB/T 17191. Bit string is enclosed in single quotes 1 and 0 strings. Such as '10000001'. The spaces in the bit string are easy to read and have no special meaning. (bitstreamleft Bitfirst). Uimsbf Unsigned integer, the most significant bit first. (unsignedinteger, mostsignificantbitfirst).

The bsmbf bit string is a quoted 1 and 0 string, with the right bit first, such as encoding a 5-bit value of 6 and then encoding a 3 ratio.

The special value is 2, then the encoded bit string is '01000110'.

4.7 Abbreviations

The following abbreviations apply to this document. FFT. Fast Fourier Transform (FastFourierTransform)

MDCT. Modified Discrete Cosine Transform (ModifiedDiscreteCosineTransform)

IMDCT. Modified Inverse Modified Cosine Transform (InverseModifiedDiscreteCosineTransform)

MDST. Modified Discrete Sine Transform (ModifiedDiscreteSineTransform)

MDFT. Modified Discrete Fourier Transform (ModifiedDiscreteFourierTransform)

IMDFT. Modified Inverse Modified Discrete Fourier Transform (InverseModifiedDiscreteFourierTransform)

MCR. Maximal CorrelationRotation

PCA. PrincipalComponentAnalysis

AASF. AVS2 Audio Storage Format (AVS2AudioStorageFormat)

AATF. AVS2 Audio Transmission Format (AVS2AudioTransportFormat)

CRC. Cyclic Redundancy Check (CylicRedundancyCheck)

BWE. High Frequency Bandwidth Extension (BandwidthExtension)

TNS. Temporal Noise Shaping (TemporalNoiseShaping)

6.1 Overview

With the application of 3Dudio (threedimensionaudio) system, 4K and 3D movies, ultra HD TV, network HD Application environments such as video, virtual reality, networking, and mobile audio have raised the need for efficient, high-quality compression coding of 3D audio data. by The amount of data in the 3D audiovisual system is much larger than that of the traditional audiovisual system, which increases the storage space and transmission bandwidth (or data traffic) overhead, so Improving the compression efficiency based on high sound quality is the main problem that AVS2 audio coding needs to solve. AVS2 audio coding is high voltage Reduced, high-quality, and audio objects are encoded as 3D audio coding techniques. Both advanced technology, rich codec options, and system integration It has high technical characteristics such as high degree of integration, flexible configuration, wide adaptability, and high performance, high compression, high sound quality and low complexity.

The AVS2 audio coding framework is shown in Figure 1. The entire system is encoded by base channel coding (base_profile) and 3D audio objects.

(3D_profile) consists of two encoding profiles. The basic channel encoding level not only combines mono, stereo (two-channel), surround sound Channel encoding technology (multi-channel) and 3D sound bed, also integrated with General Audio Coding (GA) and lossless audio coding (LosslessAudiocoding, LL) two encoding options. The universal encoding option combines two encoding modes, high bit rate and low bit rate. 3D sound The frequency object coding includes object audio data and object metadata coding, wherein the object audio data shares the basic channel coding with the 3D sound bed, and is increased.

The audio object metadata (metadata) encoding module (AudioObject Coding, Ob) realizes the type and space of the 3D audio object sound source.

The encoding of the object description information such as position and motion trajectory. Compared to current audio standards, AVS2 audio coding has five features.

a) Code stream encapsulation takes into account the storage and transmission requirements, and designs a storage format AASF (AvsAudioStorage) without frame header information.

Format) and the transmission format AATF (AvsAudioTransformFormat) with frame header and error check information. in Minimize coding redundancy while ensuring application requirements. For details, see Appendix A. The code rate difference between the two package formats is about 740b/s. b) Codec framework. […]