mpeg-1 - mcgill schulich faculty of musicich/classes/mumt621_09/presentations... · 2011. 1. 2. ·...

MPEGMPEG--11

Overview of MPEGOverview of MPEG--1 Standard1 StandardIntroduction to perceptual and entropy Introduction to perceptual and entropy codingscodings

25 September 200925 September 2009 MPEGMPEG--1 Presentation1 Presentation 22

ContentsContents

HistoryHistory

Psychoacoustics and perceptual codingPsychoacoustics and perceptual coding

Entropy codingEntropy coding

MPEGMPEG--11

Layer I/IILayer I/II

Layer III (MP3)Layer III (MP3)

Comparison and Audio QualityComparison and Audio Quality


IntroductionIntroduction

Digitizing an analog signal is (Digitizing an analog signal is (lossylossy) compression) compression

Digitizing introduces quantization noiseDigitizing introduces quantization noise

Quantization noise imply loss of qualityQuantization noise imply loss of quality

Linear quantization > 16 bit (98 dB) Linear quantization > 16 bit (98 dB) inaudible noise (CD)inaudible noise (CD)

Linear quantization 4 bit (26 dB)Linear quantization 4 bit (26 dB)


ContentsContents

HistoryHistory



MPEGMPEG--11





HistoryHistory

Moving Picture Expert Group (MPEG)Moving Picture Expert Group (MPEG)

Created in January 1988Created in January 1988

Starts the development of MPEGStarts the development of MPEG--1 in May 19881 in May 1988

Publishes the MPEGPublishes the MPEG--1 standard in November 1992 (ISO/IEC 111721 standard in November 1992 (ISO/IEC 11172-- 3 for audio)3 for audio)

MPEGMPEG--1 standard1 standard

Defines bitDefines bit--streamstream

Defines decoding functionsDefines decoding functions

DOES NOT define encoding techniquesDOES NOT define encoding techniques

Inspired by MUSICAM (Masking pattern Universal Subband Integrated Coding And Multiplexing) )


ContentsContents

HistoryHistory



MPEGMPEG--11





PsychoacousticsPsychoacoustics

Masking effectMasking effect

Critical bandsCritical bandsz/Bark lower boundary

higher boundary bandwidth

central frequency

0 0 100 100 50

1 100 200 100 150

2 200 300 100 250

3 300 400 100 350

4 400 510 110 450

5 510 630 120 570

6 630 770 140 700

7 770 920 150 840

(Brandenburg)

Time domain masking (Pohlmann 2000) Frequency domain masking (Pohlmann 2000)

Idealized critical bands (Painter & Spanias 2000)


Perceptual CodingPerceptual Coding

Dividing the different Dividing the different subbandssubbandsof a signalof a signal

Ignoring masked audio Ignoring masked audio informationinformation

Introducing inaudibleIntroducing inaudiblequantization noisequantization noise

Bits association according to masking threshold (Pohlmann 2000)

Quantization noise added according to masking threshold (Pohlmann 2000)


Perceptual CodingPerceptual Coding

Perceptual Encoder/Decoder (Kahrs & Brandenburg 1998)


ContentsContents

HistoryHistory



MPEGMPEG--11





Entropic CodingEntropic Coding

Use information about the signal to code Use information about the signal to code efficientlyefficiently

Entropy of a signalEntropy of a signal

Example 1: {0, 2, 2, 2, 0, 0, 0, 0, 0, 2, 0, 3, 2, 2, 0, 0, 0, 3Example 1: {0, 2, 2, 2, 0, 0, 0, 0, 0, 2, 0, 3, 2, 2, 0, 0, 0, 3, 0, 0}, 0, 0}20 symbols 20 symbols –– twelve 0 (0.6), zero 1 (0), six 2 (0.3), two 3 (0.1)twelve 0 (0.6), zero 1 (0), six 2 (0.3), two 3 (0.1)Entropy Entropy H H = 1.30= 1.30

Example 2: {1, 2, 3, 0, 2, 1, 1, 2, 3, 0, 0, 1, 0, 3, 3, 3, 2, 0Example 2: {1, 2, 3, 0, 2, 1, 1, 2, 3, 0, 0, 1, 0, 3, 3, 3, 2, 0, 1, 2}, 1, 2}20 symbols 20 symbols –– five 0 (0.25), five 1 (0.25), five 2 (0.25), five 3 (0.25)five 0 (0.25), five 1 (0.25), five 2 (0.25), five 3 (0.25)Entropy Entropy H H = 2= 2

Shannon theoremShannon theorem It is impossible to code with less than It is impossible to code with less than H H bits/symbolbits/symbol It is possible to code with less than It is possible to code with less than H+H+11 bits/symbolbits/symbol


Entropic CodingEntropic Coding

Huffman codingHuffman coding

Example 1: {0, 2, 2, 2, 0, 0, 0, 0, 0, 2, 0, 3, 2, 2, 0, 0, 0, 3Example 1: {0, 2, 2, 2, 0, 0, 0, 0, 0, 2, 0, 3, 2, 2, 0, 0, 0, 3, 0, 0}, 0, 0}20 symbols 20 symbols –– twelve 0 (0.6), zero 1 (0), six 2 (0.3), two 3 (0.1)twelve 0 (0.6), zero 1 (0), six 2 (0.3), two 3 (0.1)Entropy Entropy H H = 1.30= 1.30

Immediate coding:Immediate coding:0 0 ““0000”” 1 1 ““0101”” 2 2 ““1010”” 3 3 ““1111””““00101010000000000010001110100000001100000010101000000000001000111010000000110000””

Huffman coding:Huffman coding:0 0 ““00”” 1 1 ““111111”” 2 2 ““1010”” 3 3 ““110110””

““010101000000100110101000011000010101000000100110101000011000””

Efficiency:Efficiency:

Immediate coding: 2 bits/symbolImmediate coding: 2 bits/symbol

Huffman coding: 1.5 bits/symbol (statistically)Huffman coding: 1.5 bits/symbol (statistically)


ContentsContents

HistoryHistory



MPEGMPEG--11





MPEGMPEG--11

Sampling rate: 32, 44.1 and 48 kHzSampling rate: 32, 44.1 and 48 kHz

Four modes:Four modes:

Mono: 1 channelMono: 1 channel

Stereo: 2 channelsStereo: 2 channels

Dual: 2 channels independent (e.g. bilingual Dual: 2 channels independent (e.g. bilingual programmesprogrammes))

Joint stereo: 2 channels coded togetherJoint stereo: 2 channels coded together

2 perceptual models2 perceptual models

Floating point quantization (normalization)Floating point quantization (normalization)

Error checking: Error checking: Cyclic redundancy check (CRC)


ContentsContents

HistoryHistory



MPEGMPEG--11





MPEGMPEG--1 Layer I1 Layer I

From 32 to 448 kbpsFrom 32 to 448 kbps

3232--subband subband polyphasepolyphasefilterbankfilterbank

Bit allocation (0Bit allocation (0--15)15)

Max dynamic range > 120 dBMax dynamic range > 120 dB

Linear quantizationLinear quantization

1 frame 1 frame 384 samples384 samplesExample: Philips DigitalExample: Philips DigitalCompact CassetteCompact Cassette

Example of Layer I encoder (Pohlmann 2000)

Layer I frame format (Pohlmann 2000)


MPEGMPEG--1 Layer II1 Layer II


Improvement of Layer IImprovement of Layer I

Improved FFT analysisImproved FFT analysis

Scale factor redundancyScale factor redundancy

Finer quantizationFiner quantization

1 frame 1 frame 1152 samples1152 samples

Example: Digital Audio Example: Digital Audio Broadcasting (DAB)Broadcasting (DAB)

Example of Layer II encoder (Pohlmann 2000)

Layer II frame format (Pohlmann 2000)


ContentsContents

HistoryHistory



MPEGMPEG--11





MPEGMPEG--1 Layer III (MP3)1 Layer III (MP3)


Improvements:Improvements:

Finer psychoacoustics modelFiner psychoacoustics model

Alias reduction (MDCT filters)Alias reduction (MDCT filters)

NonuniformNonuniform quantizationquantization


AdaptativeAdaptative block sizeblock size

Only Layer with patentsOnly Layer with patents

Inspired by:Inspired by:

ASPEC (audio spectral perceptualentropy coding)

OCF (optimal coding in the freq. domain)

Example of Layer III encoder (Pohlmann 2000)

Layer III frame format (Pohlmann 2000)


MPEGMPEG--1 Layer III (MP3)1 Layer III (MP3)

Filtering: Hybrid Filtering: Hybrid polyphasepolyphase filter/MDCTfilter/MDCT

SteadySteady--state signals: 18state signals: 18--point MDCT on every point MDCT on every subbandsubbandFrequency resolution: 41.67 HzFrequency resolution: 41.67 HzTime resolution: 24 msTime resolution: 24 ms

Transient signals: 6Transient signals: 6--point MDCTpoint MDCTFrequency resolution: 125 HzFrequency resolution: 125 HzTime resolution: 8 msTime resolution: 8 ms

3 blocks modes3 blocks modes

PrePre--echo detectionecho detection

Quantization : power 3/4Quantization : power 3/4

Entropy coding: Entropy coding:

Huffman tablesHuffman tables

Run length codingRun length coding

Filtering stage in Layer III encoder (Pohlmann 2000)

MDCT filterbank in Layer III encoder (Pohlmann 2000)


Joint Stereo CodingJoint Stereo Coding

Intensity codingIntensity coding

Sum of left/right channelsSum of left/right channels

Coding of the sum and of left/right scale factorsCoding of the sum and of left/right scale factors

Usually only for highUsually only for high--frequency frequency subbandssubbands

Efficient for redundant audio channelsEfficient for redundant audio channels

MS (mid/side) stereo codingMS (mid/side) stereo coding

Sum and difference of left/right channelsSum and difference of left/right channels

Coding of the two valuesCoding of the two values

Stereo maskingStereo masking


ContentsContents

HistoryHistory



MPEGMPEG--11





Audio QualityAudio Quality

Comparison with CDComparison with CD--quality at 48 kHz (16 bit quality at 48 kHz (16 bit -- 1.412 Mbps)1.412 Mbps)

Layer I: Layer I:

No perceptual difference for 384 kbps (stereo) No perceptual difference for 384 kbps (stereo) –– 2:1 compression2:1 compression

Layer II:Layer II:

No perceptual difference for 256 kbps (stereo) No perceptual difference for 256 kbps (stereo) –– 4:1 compression4:1 compression

Layer IIILayer III

Increase of mean opinion score compared to Layer II at 256 kbps Increase of mean opinion score compared to Layer II at 256 kbps (stereo) for 128 kbps (stereo) (stereo) for 128 kbps (stereo) –– 8:1 compression8:1 compression


Comparison between LayersComparison between Layers

Layers I and II are very similarLayers I and II are very similar

Each Layer has its defined decoder design Each Layer has its defined decoder design

Encoding/Decoding complexity: Layer I/IIEncoding/Decoding complexity: Layer I/II broadcastingbroadcasting

Encoding/Decoding quality: Layer IIIEncoding/Decoding quality: Layer III audio storageaudio storage


ConclusionConclusion

BibliographyBibliography::

AmbikairajahAmbikairajah, E. & al. , E. & al. Auditory masking and MPEGAuditory masking and MPEG--1 audio compression1 audio compression, , Electronics & Communication Engineering Journal, Electronics & Communication Engineering Journal, 19971997

Brandenburg, K. & Brandenburg, K. & BosiBosi, M. , M. Overview of MPEG Audio: Current and Future Overview of MPEG Audio: Current and Future Standards for LowStandards for Low--BitBit--Rate Audio CodingRate Audio Coding, Journal of the Audio Engineering Society, , Journal of the Audio Engineering Society, 19971997, Vol. Vol. 45(No. 1/2) , Vol. Vol. 45(No. 1/2)

Painter, T. & Painter, T. & SpaniasSpanias, A. , A. Perceptual Coding of Digital AudioPerceptual Coding of Digital Audio, Proceedings of IEEE, Proceedings of IEEE, , 20002000, Vol. Vol. 88(No. 4), Vol. Vol. 88(No. 4)

Painter, T. & Painter, T. & SpaniasSpanias, A. , A. A Review of Algorithms for Perceptual Coding of Digital A Review of Algorithms for Perceptual Coding of Digital Audio SignalsAudio Signals,, Digital SignalDigital Signal Processing, Processing, 19971997

Pan, D. Pan, D. A Tutorial on MPEG/Audio CompressionA Tutorial on MPEG/Audio Compression, IEEE , IEEE MultiMediaMultiMedia, IEEE , IEEE Computer Society, Computer Society, 19951995, Vol. 2(2), pp. 60, Vol. 2(2), pp. 60--7474

Pan, D.Y. Pan, D.Y. Digital Audio CompressionDigital Audio Compression, Digital Technical Journal, , Digital Technical Journal, 19931993, Vol. 5, Vol. 5

KahrsKahrs, M. and Brandenburg, K. , M. and Brandenburg, K. Applications of digital signal processing to audio Applications of digital signal processing to audio and acousticsand acoustics, , KluwerKluwer Academic Publishers, Academic Publishers, 19981998

MallatMallat, S. , S. TraitementTraitement du Signaldu Signal, , EcoleEcole Polytechnique, Polytechnique, 20002000

PohlmannPohlmann, K.C. , K.C. Principles of Digital Audio,Principles of Digital Audio, McGrawMcGraw--Hill Professional, Hill Professional, 20002000

http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=620475&isnumber=13490http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=620475&isnumber=13490http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=620475&isnumber=13490http://www.aes.org/tmpFiles/elib/20090924/7871.pdfhttp://www.aes.org/tmpFiles/elib/20090924/7871.pdfhttp://www.aes.org/tmpFiles/elib/20090924/7871.pdfhttp://www.aes.org/tmpFiles/elib/20090924/7871.pdfhttp://www.aes.org/tmpFiles/elib/20090924/7871.pdfhttp://www.aes.org/tmpFiles/elib/20090924/7871.pdfhttp://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=842996&isnumber=18261http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=842996&isnumber=18261http://medialab.hsr.ch/dm/vorlesung/dsp97.pdfhttp://medialab.hsr.ch/dm/vorlesung/dsp97.pdfhttp://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=00388209http://eceftp.niu.edu.tw/mhyeh/teach.files/media92/data/paper/Digital_Audio_Compression_1993.pdfhttp://www.springerlink.com/content/h77t14/?p=e0f426fe6ce54d0cbf51591d35ae5987&pi=0http://www.springerlink.com/content/h77t14/?p=e0f426fe6ce54d0cbf51591d35ae5987&pi=0

MPEG-1ContentsIntroductionContentsHistoryContentsPsychoacousticsPerceptual CodingPerceptual CodingContentsEntropic CodingEntropic CodingContentsMPEG-1ContentsMPEG-1 Layer IMPEG-1 Layer IIContentsMPEG-1 Layer III (MP3)MPEG-1 Layer III (MP3)Joint Stereo CodingContentsAudio QualityComparison between LayersConclusion

mpeg-1 - mcgill schulich faculty of musicich/classes/mumt621_09/presentations... · 2011. 1. 2. ·...

Documents