Audio & Video Compression and its Application in Consumer Products

Alain Bouffioux - Philips Leuven 1A/V compression & Consumer Products - ULG

Philips Digital Systems Laboratories 09/12/2004

Audio & Video Compressionand its Application inConsumer Products



Agenda• Introduction - The evolution of Audio/Video

consumer products and the role of compression techniques.

• Audio & Video compression principles

• Audio compression

• Video compression

• Audio/Video synchronisation

• The MPEG model and its situation in a communication context

• Application to DVD (Digital Versatile Disc)

• Application to DVB (Digital Video Broadcasting)

• Conclusion



• Introduction - The evolution of Audio/Video consumer products and the role of compression techniques.








• Conclusion

Agenda



Moore’s law

Number of transistors per square inch doubles every 18 months



Moore’s law today

Cost of a transistor divided by one million in 30 years



Moore’s law today (2)

• “Self-fulfulling prophecy” (not automatic)

= roadmap for the semiconductor industry



Moore’s law today (3)• Roadmap for semiconductor industry

= only certainty in the current undefined future• Moore’s law will continue to apply: 20 years

– Economical limitation ?– Power consumption (Moore’s low in reverse direction)– Architectural gap between IP-blocks & application

(middleware still more complex…)• Progresses in semiconductors

= fuel the innovation = fuel the software revolution= fuel the wireless revolution

(WLAN, WPAN, WBAN, …) • …• Examples:

WBAN & sensors, RFID applications, camera to swallow, flexible display…



The evolution of some CE products (1)

ConsumerConsumerComputerComputer

CommunicationCommunication

CDCDDCCDCC

CD-iCD-i

DVDDVD

STBSTB



The evolution of our CE products (2)• The Residential Gateway (Set-Top-Box) as the link

between the home and the world-wide information infrastructure.

STB Home NetworkWorld-wide

communicationinfrastructure



The evolution of our CE products (3)

• The STB (in home) as the gateway to various services. Local Server provides 2 kind of services:– Broadcast

Analogue & digital TV, NVOD, PPV– Point-to-point (Home to local server)

Home shopping, VOD, e-mail, Web browsing, PC connection...

Network

Local server

Internet

Local server

Up to 800 homes



The evolution of our CE products (4)• The STB as a key element of the home network

Home Network

Residential Gateway

To telephone Network

To satelliteNetwork

To cable Network

Computer

DVD Jukebox

Television

DiskRecorder



The evolution of our CE products (5)

• 3C Convergence - Progressive

• New products combine all 3 functions

• Products always more and more complex

• Products have always new features

• Lifetime of products is always shorter



Compression is one among the various factors (all powered by semiconductor progresses) that enable multimedia.

Software (methodology,user interface ...)

International cooperation(interoperability &economy of scale)

Disc capacity (DVD), communicationgoing digital (Modems, xDSL, ATM ...)

Multimedia

Electronics(Memory capacity,clock frequency,µP architecture, ...--> decoding at low cost

Audio/VideoCompression(e.g. CCIR601 vs MPEG)

Factors enabling such evolution



BUT !!• Convergence of technologies (consumer,

communication, computer) All products combine all three technologies

• BUT !

• Divergence of applications– Home consumer, Multimedia phone, Camera, PDA,

Office computer, Automotive…

– High number of potential productsTechnology push Market pull (user centric approach)












• Conclusion



Compression in first A/V Products (1)• First Audio/Video products made compression

without knowing it was compression.

• How ?By removal of irrelevancies

• Audio and Video characteristics

Audio VideoSpectral

SensitivityGood Bad

SpatialSensitivity

Bad Good



Compression in first A/V Products (2)• Audio products

From 2 to 7.1 channels are enough to provide the spatial resolution.

• Video productsThree colours (RGB) are enough to provide the spectral resolution.



• Audio: Compression needed in spectral domain

• Bitrate of a stereo audio source (CD-DA encoding)

Sampling frequency : 44.1 kHzStereo16-bit per sampleBitrate = 44100 * 2 * 16 = 1.41 Mbit/sec

Audio waveform (time)

time

The need for more compression (1/5)



• Video: Compression needed in spatial domain

• Bitrate of a video source (CCIR 601 - 50 Hz countries)

25 images per secondYUV coding (Y: luminance - U,V : Chrominance)Y: 8 bit per pixel - U,V: 1 pixel on 2 coded, 8 bit per pixelBitrate = (576*720)*25*16 = 166 Mbit/sec

The need for more compression (2/5)

720 samples

576lines

Video image



The need for more compression (3/5)• Channels availables for AV transmission

– Analog television channel (compatibility)Cable (bandwidth = 8 MHz) Satellite (Bandwidth = 30-40 MHz)

Capacity around 40 Mbit/sec

– Compact disc (CD)For 74 min. play time : 1.41 Mbit/sec



The need for more compression (4/5)• MPEG-1 target (Moving Picture Expert Group)

(Video-CD : 74 min. constraints)

But quality was judged too poor (about VHS quality)

Compression

Video : 166 Mbit/sec

Audio : 1.4 Mbit/sec

1.4 Mbit/sec



The need for more compression (5/5)• MPEG-2 target

– Program stream (DVD)

– Transport stream (DVB)

Compression

1 program(video, multichannelaudio, ....)

= motivation for the capacityincrease of the CD (--> DVD)

3-9 Mbit/sec (variable bitrate)(but higher quality than MPEG-1)

Compression

n programs(video, multichannelaudio, ....)

about 40 Mbit/sec (constant bitrate)(DVB-Satellite & DVB-Cable)



Principles of compression (1/2)• Compression (or source coding) is achieved by

suppressing information : – redundant information – irrelevant information

• Suppression of redundant information lossless compression example: PCM to DPCM,DCT

The original signal and the one obtained after encoding and decoding are identical

DecompressionCompressionFc(x,y,t)

Rc kbps Ri < Rc

Fp(x,y,t) = Fc(x,y,t)

Rp = Rc



Principles of compression (2/2)• Suppression of irrelevant information

lossy compression Example: bandwidth limitation, masking in audio

•

The original signal and the one obtained after encoding and decoding are different but are perceived as identical

DecompressionCompressionFc(x,y,t)

Rc kbps Ri < Rc Rp = Rc

Fp(x,y,t) <> Fc(x,y,t)



Audio Demonstration

From “Borderline” Madonna - Stereo - 16 bit/channel

Compression used AAC

Compression

Decompression

Original

-

32kbps

128kbps

64kbps

16kbps

705 kbps



MOS scale (1/2)• Signal distortion is not a good measure of the

performance of a lossy compression method an other method is necessary: MOS scale (Mean Opinion Score)

• The five-grade CCIR impairment scale (Rec.562)1(Very annoying), 2(Annoying), 3(Slightly annoying), 4(Perceptible but not annoying), 5(Imperceptible)

• Example:Double blind test



MOS scale (2/2)

Compressed

3

1

4

Impairmentscale5

Min value

Sequence2 3

Original

Mean value

Max value

Original signal

Listener answers to :1. Which signal is the original ? 2 or 3 ?2. Grade the other one ?

Exchange box(Random for listener)

Selector(controlled by listener)

Compressed signal



Compression to VBR or CBR

• CBR (Constant Bit Rate) vs VBR (Variable Bit Rate)

• Scene more complex Higher bit rate for same quality

• CBR variable quality (example : Video CD artefact)

• Constant quality VBR necessary (e.g.: DVD-Video)

Constantbit rate

Bitrate

Complex

Distorsion

Simple

Constant quality



Video demonstration



The compression trade-off• Compression techniques are still making progress

• Trade-off Complexity/Quality/Bit Rate

• New technique may result in new trade-off

QualityQuality

BitrateBitrate

ComplexityComplexity

MPEG Layer 1MPEG Layer 1



MPEG AACMPEG AACOther TechniqueOther TechniqueSpeech codingSpeech coding












• Conclusion



Audio compression in MPEG (1/5)• Based on psycho-acoustics

• Compress the bit rate without affecting the quality perceived by the human ears (based on the imperfection of human ears)

• Removal of irrelevancies

• 4 main principles :– Threshold of audibility

– Frequency masking

– Critical bands

– Temporal masking



Audio compression in MPEG (2/5)• Principle 1: Threshold of audibility

Not all frequency components need to be encoded with the same resolution. Nr_bit(f) = (signal/threshold)db/6

0.2

Frequency (kHz)

20

0

0.02 0.10.05

100

80

60

40

Sound level(db)

50.5 1 2 2010



Audio compression in MPEG (3/5)• Principle 2: Frequency masking

Analysis of the incoming signal

0.2

Frequency (kHz)

0

20

0.050.02 0.1

60

40

80

Sound level(db) 100

Masked signal

0.5 1 2 5 10 20

New threshold of audibilitydue to masker signal

Masker signal



Audio compression in MPEG (4/5)• Principle 3: Critical bands

– Human ear may be modelled as a collection of narrow band filters

– Bandwidth of these filters = critical band

– critical band(<100 Hz) for lowest audible frequencies( 4 kHz) for highest audible frequencies

– The human ear cannot distinguish between two sounds having two different frequencies in a critical band.Example : when we hear 50 & 60 Hz at the same time we cannot distinguish them.

– Consequence : Noise masking threshold depends solely of the signal energy within a limited bandwidth domain.The largest sound is taken as the representative of the critical band.Necessity to analyse the signal at 100Hz resolution at low-frequency



Audio compression in MPEG (5/5)• Principle 4: temporal masking

selection of the frame duration for frequency analysis and encoding.

0~(-10msec) 0

Envelope of masker

Pre masking

Level(db)

Post masking

Simultaneous masking

Time~100msec



An enabling tool : the filter bank (1/2)

Digital audio signal, Fs

ENCODING / DECODING

Reconstructed signal, Fs

Synthesis

INTERPOLATION

Analysis

DECIMATION

n subband signals, Fs

Filter bank

n subband signals, Fs

n subband signals, Fs/n

n subband signals, Fs/n

Filter bank



An enabling tool : the filter bank (2/2)• After decimation, same bit rate as original signal, but

signal decomposed in various frequency ranges possibility of frequency based compression

• Filter-bank:Aliasing occurs due to decimation

• It exists a class of filter-bank such that aliasing is compensated in synthesis filter : QMF (Quadrature Mirror Filter) but high complexity

• Pseudo-QMF (Polyphase filter bank) is used. Has good compromise between computation cost and performances

• Remark : Aliasing may occur if signal in a adjacent band is not reconstructed with an adequate resolution.



The MPEG encoder

Signal to mask level for each subband(quantisation information)

Psychoacousticmodel.

Digital audioinput

Subbandsamples

FilterBank

Ancillary data

Quantisedsamples

Bitallocation

Bitstreamformatting

Encodedbitstream



The MPEG filter bank • In MPEG, 32 equal-width subbands are used

• For each subband, necessity to define the maximum signal level and the minimum mask level.

• BUT, at low frequencies : bandwidth of subbands > critical bands

• Necessity to rely on an FFT in order to compensate the lack of frequency selectivity of filterbank at low frequencies



Psychoacoustic model & Bit allocation(1/2)

• An FFT compensates the lack of frequency selectivity of filterbank at low frequencies

• FFT : 512 samples (layer 1) & 1024 samples (layer 2)resolution for layer 1 : Fs/512 < 100 Hz

• A psychoacoustic model based on the FFT computes the signal to mask ratio for each subband (1 bit = 6db)

• Ideally, after allocation, quantisation noise < masking level

• The scale factors are computed for each subband from the filterbank output (floating point representation of samples)

• The bit allocator adjust the bit allocation in order to meet the bitrate requirement.

• The bitstream syntax is dependent of the MPEG layer (See later)



Psychoacoustic model & Bit allocation(2/2)

Frequency

Signal/MaskLevel(db)

Level(db)

Signal level

Bit allocation= 1 bit

Frequency

Mask level



The MPEG decoder• Decoder is simple (Complexity is at encoder side)

• Remark 1: DCC is MPEG-1 but DCC encoder has no FFT, relies only on power in the 32 subbands Higher bit rate (320 kbps) to reach transparent quality

• Remark 2: MPEG specifies bitstream syntax only. Encoder are given for information. Possibility of improvement.

SubbandSamplesrecons- truction

Bitstreamunpacking

Encodedbitstream

Ancillarydata

Subbandsamples

Filterbank

Decoded audio signal



Audio features in MPEG• MPEG1 :

– Mono/stereo/dual/joint stereo (Possibility Dolby surround)

– Sampling frequencies : 32, 44.1 & 48 kHz– 3 layers : trade-off complexity/delay versus

coding efficiency of compression– Various bit rate : trade-off quality versus bitrate

• MPEG2 :– 5.1 channels– Sampling frequencies extended to 16, 22.05 &

24 kHz



Dolby surround principles (1/5)• 4 channels carried by stereo pair same tools as for stereo

• Compatible with stereo installation

+90°

-90°

Phaseshifter

Lt

RtSurround Surround

Configuration

Center

Center(C)

- 3 dB- 3 dB

Right (R)

Surround(S) 100 Hz -

7000 Hz

BP

- 3 dB- 3 dB

From 4 channels to 2 channels

Left (L)



Dolby surround principles (2/5)

E n c o d i n g e q u a t i o n s S i m p l e d e c o d e r

Lt L .1

2

C ..j1

2

S L Lt Ld L .1

2

C ..j1

2

S

Rt R .1

2

C ..j1

2

S R Rt Rd R .1

2

C ..j1

2

S

CRt Lt

2

Cd .1

2

L .1

2

R C

SLt Rt

.2 j

Sd ..j1

2

L ..j1

2

R S



Dolby surround principles (3/5)

3 dBS

3 dB

3 dB

L C

3 dB

R

• Simple decoder provides only 3 dB channel separation(See previous equations) Need for improvement Dolby Surround pro-logic decoder (next slide)



Dolby surround principles (4/5)Dolby surround pro-logic decoder

BP

C'

S'

R'

L'

Rt

-3 dB

Lt

-3 dB Direc-tionCom-pen-sation-90°

Rd

Sd

VCA

VCA

Cd

Ld

VCA

VCA



Dolby surround principles (5/5)Performance of Dolby pro-logic decoder

Channel separation larger than 35 dB

> 35 dB

S

> 35 dB

> 35 dB

L

> 40 dB

C

> 35 dB

> 40 dB

R



5.1 surround soundMPEG-2 surround configurations (front/back)

• 3/2

• 3/0 + 2/0

• 3/1

• 2/2

• 2/0 + 2/0

• 3/0

• 2/1

• 2/0

• 1/0

+ LFE (opt.) (Fs/96) 15-120Hz

SurroundLeft

SurroundRight

(+LFE)

Center Right



Virtualisation• Virtualisation has no direct relation with the MPEG

standard.It is considered here only because it may be implemented in some of the future audio products (DVD, STB ...)

• Virtualisation is a product feature.

• It allows reproduction of surround information (5.1, 3/1) on a stereo installation.



Virtualisation principle

• Virtualisation = processing of the signal in such a way the source of the signal is perceived at a selected position outside the loudspeaker axis (virtual loudspeaker).

• Drawback : very sensitive to listener position (stability)

• Remark : a mono signal coded in normal stereo is perceived between the two loudspeakers

Listener

Real loudspeakerProcessing

Mono signal

Virtualloudspeaker



Stereo widening

• Also called Q-sound , incredible sound, azimuth positionning ...

• The stereo sources are positionned at virtual locations for improving the stereo effect (cheap analog solution exists)

• Real sound comes from real loudspeakers. Perceived sound is as if stereo signals were coming from virtual loudspeakers

Listener

Real loudspeakerProcessing

Stereo signal

Virtualloudspeaker



Virtual surround• Virtual surround gives on a stereo installation the

subjective effect of a multichannel configuration.

• Each channels is virtually positionned at a location around the listener. The stereo installation performs the addition of the processed signals for each audio channel.

• Real sound comes from a stereo installation. Perceived sound is as if the various surround signals were coming from some virtually located loudspeakers.



Summary of surround aspects

Remarks about Dolby surround pro-logic :

Only carrier is stereo, source & presentation are multichannel

Compatible with stereo installation (no surround effect except in the case of surround virtualisation)

Stereo receiver Multichannel receiver

Stereo source Stereo widening(incredible sound,azimuth positioning)

3-D special effects(hall, theatre, stadium)

multichannel source Virtual surround Multichannel output












• Conclusion



Video compression in MPEG (1/6)• Principles

– removal of intra-picture redundancy : Image is decomposed in 8*8 pixels sub-images.Each sub-image contains redundant information DCT transformation (in frequency domain) decorrelates the input signal.( most energy in low spatial frequencies)

– removal of interpicture redundancy :coding of difference with an interpolated picture (moving vectors)

– high frequent spatial frequencies quantized with lower resolution than low ones(remove irrelevancy)

– zig-zag scan and VLC (remove redundancy)



Video compression in MPEG (2/6)• Result

– 4:2:2 CCIR 601 resolution : 166 Mbps (=25images/sec *576lines* 720pixels* 2(lum & chrom) *8bits) ± 3-4 Mbps (mean) in MPEG2

– 4:2:0 SIF resolution : 30 Mbps (=25 images/sec *288 lines *352pixels* 1.5(lum & chrom) *8bits) ±1.2 Mbps (CBR) in video CD (MPEG1)



Video compression in MPEG (3/6)• Spatial redundancy reduction (DCT example)

158 0 -1 0 0 0 0 0 -1 -1 0 0 0 0 0 0 -1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

139 144 149 153 155 155 155 155144 151 153 156 159 156 156 156150 155 160 163 158 156 156 156159 161 162 160 160 159 159 159159 160 161 162 162 155 155 155161 161 161 161 160 157 157 157162 162 161 163 162 157 157 157162 162 161 161 163 158 158 158

158 0 -1 -1 -1 -1 EOBzig-zag scan

1260 -1 -12 -5 2 -2 -3 1 -23 -17 -6 -3 -3 0 0 -1 -11 -9 -2 2 0 -1 -1 0 -7 -2 0 1 1 0 0 0 -1 -1 1 2 0 -1 1 1 2 0 2 0 -1 1 1 -1 -1 0 0 -1 0 2 1 -1 -3 2 -4 -2 2 1 -1 0

DCT

Quantisation



Video compression in MPEG (4/6)• Temporal redundancy reduction

B

5

Bi-directional prediction

I : Intra-coded pictureP: Predicted pictureB: Bi-directionally interpolated picture

4

B

Order ofpresentation

Order oftransmission

BI P

0 3

B P

1 2 6

B

Prediction

I B P B

Increase of compressionrate

0 1 2 3 4

7

B P B

5 9

B I P

8

P B B P I B

86 7 9



Video compression in MPEG (5/6)• Model of a possible encoder

Buffer

Data

Regulator

Motion estimation

Motionvectorsand modes

Format conversion and picture reorder

Inputpictures

Picturestore andprediction

1/DCT

DCT VLC

1/Q

Q Multi-plex



Video compression in MPEG (6/6)• MPEG1 en MPEG2 video features

– MPEG1

• sequential picture

• resolution : SIF format 288(240)*356*24,25 or 30 Hz

– MPEG2

• sequential or interlaced

• various levels : low level (SIF: 288*356), main level (CCIR601: 576 * 720), high 1440 level (HDTV: 1152*1440), high level (EQTV: 1152*1920)

• various profiles (toolboxes) : simple profile (No B picture), main profile (=MPEG1+interlaced), SNR scalable profile (allows graceful degradation (noise improvement at same resolution), spatial scalable profile (hierarchical coding : improvement at higher resolution), high profile.












• Conclusion



Synchronisation

Synchronisation in the multimedia context

refers to the mechanism that ensures a temporal

consistent presentation of the audio-visual

information to the user



Intramedia synchronisation

T between capture & presentation = Constant

Same clock frequency & Data on time Need for corresponding tools

Network

T1T2 =

T1

Capture time

Audio signal Encoder

T = Constant

Decoder

Presentation time

Audio signal



Intermedia synchronisation

T_Audio = T_Video Sampled at the same time Presented at the same time) Possible tools : common time base and presentation control (media

synchronisation with the common time base)

Ex.: Lip_sync (requirement: |delay_difference| < 80msec)

Network

T_audio = Constant

T_video = Constant = Capture time

Video signal

Audio signal

Encoder

Capture time

T_Audio Presentation time

Decoder

Video signal

Audio signal

Presentation time



Recovery of clock in CBR• CBR = Constant Bit Rate

• if the clock to recover is synchronous with transport clock Recovery of clock but not of common time base

• Remark : possibility to slave DSM to local clock

Filter

CBR stream

Phaseerror

Filling level

50%

Time Information carriedby each sample

VCO

Processing

Recoveredclock

time



Recovery of clock and time base in VBR

• VBR = Variable Bit Rate • Need for insertion of time stamps (OUTPUT TIME)

Output time stamp says for example : “It is now 16h25”Receiver adjusts its own horloge to the received time stamp

• Recovery of clock & of common time base

Time information carriedonly by time stamps

Time stamp extraction

Time counter

Recovery of clock & time

First time stamp

Others

Clock

Data stream

Data stream

Time stamps

Counter

Recovered clock

Error Filter VCO

VBR Stream

Counter sample(=Time stamps)

Channel

Time stamps



Synchronisation with common time base• Insertion of time stamp (=INPUT TIME)

Input time stamp says : “Input has been sampled at 16h29”.Receiver presents the sample at (its input time stamp + maximum encoding and decoding delay).Alternative: transmission of presentation time stamp (input time+delay)

Buffering

Mediaoutput

Media input

Time clock(Recovered)

Comparison of time clockwith sampled time clock

Sample "Time clock"Assemble frame

Time clock

Timestamp

Processing

Processing

Channel



Getting data on time• “On time” Not too late, not too early

No buffer over- or underflow

• Flow control : not applicable in broadcasting

• Common time base and Definition of a standard target decoder that describes the data consumption pattern of the receiver.

• Remark: Direct MPEG (Microsoft) does not use time information for clock recovery but relies on flow control



Streams• Idea of continuity (pipelining)

• Carry time information for clock recovery

• No flow control (allows broadcasting)The emitter must have a precise knowledge of the receiver data consumption pattern (explicit in MPEG STD)

• Just-in-timeShorter delay and smaller buffer size than with flow control

• Two aspects in synchronisation :Clock recovery & timing control (model & buffering)



Requirement on for stream transport

• Data information BER (Bit Error Rate) requirementNo repetition of frame possible FEC (Forward Error Correction)

• Time information No jitter












• Conclusion



What is MPEG ? (1/2)

• Moving Picture Expert Group

• Still active (MPEG-21 is currently in development)

• International standard (ISO/IEC) Interoperability & economy of scale

• Compression of audio and video and multiplexing in a single stream

• Definition of the interface not of the codecs room for improvement

• MPEG-1 : until 1.5 Mbps, for DSMProgressive picture, stereo (Dolby surround)



What is MPEG ? (2/2)• MPEG-2 : Various bit rates (CBR & VBR)

Program stream for DSM, transport stream for networkInterlaced picture, 5.1 audio channels Definition of various video levels (e.g. CCIR601 resolution: 4-9 Mbps, HDTV:15-25 Mbps) and profiles

• MPEG-3 : Cancelled, integrated in MPEG-2(Initially : for HDTV)

• MPEG-4 : standard for audio, video and graphics in interactive 2D and 3D multimedia communication. (Initially : low bit rate for real-time personal communication)

• MPEG-7 : Multimedia contents description interface

• MPEG-21 : Focus on multimedia distribution and on DRM aspects.



The MPEG model (1/2)

Audiodecoder

Audio signal

Videosignal

Presented signals

Multiplexer

Videodecoder

Captured signals

Audioencoder

Videoencoder

Audio signal

Videosignal Digital storage medium

orNetwork

Transmission channel

Demulti-plexer



The MPEG model (2/2)• Compression of audio & video and multiplexing in a single

stream

• Guarantees intramedia and intermedia synchronisation.

• MPEG defines an interface– bitstream syntax

– timing of the bitstream STD specifying timing requirement (ideal model)

• Consequences:– Decoder should compensate deviations from STD

– Network should correct jitter introduced by the channel (RTD-LJ)

• MPEG stream must be adapted to transmission channel formatting, error correction, channel coding (b.v.video-CD)



Components of the MPEG standard• The MPEG standard is composed of 3 main parts :

– Audio : Specifies the compression of audio signals

– Video : Specifies the compression of video signals

– System : specifies how the compressed audio and video signals are combined in the multiplexed stream (program stream or transport stream).

• Each part specifies :– The bitstream syntax

– The timing requirement and the related information (bit rate, buffer needs)



Synchronisation Mechanism (1/2)

System decoderanddemultiplexer

Multiplexer and systemencoder

Audiodecoder

Videodecoder

ComparisonPTS and STCand presentation

Videooutput

ComparisonPTS and STCand presentation

Audiooutput

Extractionof PCR (SCR)

STC

Audioencoder

Videoencoder

Assemble pictures,Sample STC for PTS

Videoinput

Assembleaudio frames,Sample STCfor PTS

Audioinput

Sample STCfor PCR(SCR)

STC

Transmission channel



Synchronisation Mechanism (2/2)• PCR for TS & SCR for PS (but same concept)

• Clock & time base recovery: Time-stamping at OUTPUT (PCR included in TS multiplex, SCR in pack header)

• Audio & video clock locked to STC easy recovery (see next slide)

• Synchronisation of audio & video to common time base (Time stamping at Input)

• STD is defined (because of the absence of flow control)streams are such that STD buffers never over- or underflow

• In TS, many program in a single stream but unique clock per program.

• Time information “No Jitter” requirement for transport



Clock recovery in receiver

VCO Audio clockdivider

Video clockdivider Video

clock

Audioclock

Audiooutput

STC

ComparisonPTS and STCandPresentationDecoded

audio

PTS

STC(Counter)

PCR

STC

Error Low Pass Filter(Integrator)

Load first PCR



MPEG program & transport streams• Program streams:

– Relatively error free environment

– program stream packet may have variable and great length

– Single time base

• Transport streams:– environment where errors are likely

– many programs (independent time base)

– Transport stream packet : fixed, 188 bytes

– Contains tables



Channelencode

Channeldecode

Bit-stream

Sourcedecode

Informationsink

Digitaloutput

Format Decrypt

SourceencodeFormat

Informationsource

Digitalinput

Encrypt

Synchro-nisation

Digitalwaveform

Otherdestination

Demodu-late

Demulti-plex

Multipleaccess

Modu-late

Multi-plex

Channelbits

Othersources

Multipleaccess

Channel

MPEG in a communication context (1)• “Typical” communication system



MPEG in a communication context (2)• MPEG : Source coding only (bit rate reduction) +

multiplexing

• The MPEG stream must be adapted to the channel in what concern its physical characteristics and in order to get the required QoS (Quality of Service) & Security

– Encryption

– Channel coding (forward error correction, interleaving, modulation codes)

– multiplexing & formatting

– modulation (frequency allocation)

– multiple access method

• Some channels : CD/DVD - satellite - cable - ATM - 1394



VideoEncoder

MPEG2 compression layer

Audioencoder

Audio,videosources

ES(ElementaryStream)

Adap-tationto thechannel

PS(1 pro-gram)

MPEG2 system layer

PSMulti-plexing

Adap-tationto thechannel

DVB, DVD ...

Disc

Satellite

TSMulti-plexing

TS(n pro-grams)

Adap-tationto thechannel Cable

TS (Transport Stream)orPS (Program Stream)

MPEG in a communication context (3) • A simple view of MPEG in the communication

context



• Introduction - The evolution of Audio/Video consumer products and the role of compression techniques.








• Conclusion

Agenda



CD : Some concepts

• Hard disk vs compact disc : more differences than just storage technique.HD developed for data storage and recording, CD developed for stream storage (CD-DA) their basic differences

• Questions– track form? – read direction? Why?– CAV or CLV? Why?– Access time : CD-ROM vs HD?– Data storage: on which face? – Production method?– Capacity?– Sensitivity to error? Diameter of a possible hole?– Standard = Interface definition : CD vs HD ?



CD-DA: Encoder model (1/3)

Modulation

6 samples = 24 bytes= 1 frame

EFM + 3 merging bits

Synchronisation pattern27 bit/frame

561 bits / frame

CD-ROM1 sector = 98 frames75 sectors/sec.

588 channel bit/frame

Right

LeftA/D conversion

PCM 44.1 kHz16 bit/sample/channel

Subcode(1 byte / frame)

Physical layer

Error correctionencoding

32 bytes/frame



CD-DA: Encoder model (2/3)• The CD-DA physical layer adapts the input stream (audio) to the

requirements of the channel

– Modulation : EFM (Eight to fourteen modulation + 3 merging bits) Pit & land length (number of successive 0 or 1 as written to disc): between 3 and 11 channel bits DC free code for adaptation to the channel bandwidth & for clock recovery considerations.

– Error correction (Cross-interleaved Reed-Solomon code)Interleave placed between C1 & C2 ECC.Next slide presents only principles and not real CD implementation.



– Error correction : addition of redundancy in order to be able to correct errors (e.g. RS(28,24,5)*RS(32,28,5))Principle :

– Interleaving : time diversity in order to deal with error burst.Successive erroneous channel bits (burst error) do not damage the same Reed-Solomon table.

CD-DA: Encoder model (3/3)

4 bytesC2 codewords

24 bytes

28 bytes

Data

4 bytes

C1code-words



CD-ROM encoder model

2336 bytes

Mode 1

Mode 2

Mode 1

2336 bytes

Additionalerror detection &error correctionencoding

Optional EDC 4 bytes

User data 2324 bytes

Subheader 8 bytes

Subheader 8 bytes


Sync Pattern 12 bytes

Additionalerror detection &error correctionencoding


Zeroes 8 bytes

Header 4 bytes

Header 4 bytes

2340 bytes

For EDC only

Video-CD uses CD-ROM mode 2 sectors

OR

12 bytes

2340 bytesScramblingOR1 CD-ROM sector= 2352 bytes



From CD to DVD : the motivation• Motivation = increase the capacity

• Why ? - Requirement of the motion picture industry

– Playback time : more than 135 min. (duration of 90% of films)

– Picture quality : superior to laser disc– Audio quality : 5.1 channels surround– Language/subtitles : 3 languages minimum.

capacity needs : more than 4.7 Gbytes• Where ? - In physical layer

• DVD : developed specifically for audio/video( video CD).



The DVD physical layer (1/2)

Recorded sectors

2366 bytes(13 *182 bytes)

8/16modu-lation

37856 channel bits

Synchronisation(2*13)*32 channel bits

38688 channel bits(eqv. to 2418 bytes)

EDC : Error Detection CodeECC : Error Correction CodeCPR-MAI : CoPyRight MAnagement Information

CPR-MAI - 6 bytes

ID (incl. sector#)4 bytes

Data2048 bytes

+EDC

Scrambling

EDC4 bytes

6 bytes

ECC(per group of 16 sectors)

Data sector

2064 bytes(12 * 172 bytes)

RowInter-lea-ving



• Objective was the storage of 2K sectors

• Error Correction Code (Reed-Solomon) - add redundancy

• Modulation - time diversity(Number of consecutive 0 : between 2 and 10)Pit and land length : between 3 and 11 (Idem CD)

• Synchronisation : for sector reconstruction.

The DVD physical layer (2/2)

16 bytesPO (Outer Parity)

172 bytes

192 bytes

16 data sectors

(12 * 172 bytes)*16

10 bytes

PI(InnerParity)



DVD: the capacity improvement (1/4)

• Increase of channel bit density ( gain = 4.50)Min pit length : (0.83 0.4)Track pitch : (1.6 0.74)Diameter of laser spot ( wavelength/NA)Wavelength (780 640 nm) gain = 1.5NA (0.45 0.60) gain = 1.78reduced margin gain = 1.68

• Modulation:EFM (8 to 17 bit) 8 to 16 gain = 1.06

• Error correctionRS(32,28,5)*RS(28,24,5) RS(182,172,11)*RS(208,192,17)

gain = 1,16




• No subcode gain = 1.03

• Sync pattern gain = 1.03

• Better sector formattingsector length (2352 bytes 2064)gain = 1.14

• Other (e.g. recorded area) gain = 1.07

Total gain : 7.2

Capacity per side : 650 MBytes (mode 1) 4.7 Gbytes




0.6 mm

0.6 mm

0.6 mm

0.6 mm

For layers 0 and 1

A side

Single-layer double-sided disc

Single-layer single-sided disc

0.6 mm

0.6 mm

B side

Dual-layer double-sided disc

0.6 mm

0.6 mm

Dual-layer single-sided disc



DVD: the capacity improvement (4/4)• Capacity of the various types

Single-layer single-side 4.7 GbytesDual-layer single-side 8.5 GbytesSingle-layer double-side 9.4 GbytesDual-layer double-side 17 Gbytes



Part 1 : Physical specification

Part 2 : File system specification

Part 3 : Video specification

How blocks may be retrieved. Definition of the file andvolume structure.

How blocks of 2048 bytesare stored on the disc

Contents of the data block.How audio and video are mapped to the block, file and volume structure

The 3 components of the DVD-V standard• DVD = DVD (= 3 random letters) (previously :

Digital Versatile Disc, Digital Video Disc)

• DVD-V : DVD - Video



Some DVD-V features (1/2)

Presentation data = MPEG program stream, VBR, max peak bit rate = 10.08 Mbps)

Video data 1 stream Mpeg1Mpeg2 (ML@MP)16:9 or 4:3 aspect ratioNTSC or PAL

Audio data max 8 streams Mpeg2 + 7.1 extension(50 Hz countries)AC-3 (60 Hz countries)Linear PCM (incl. 96 kHz- 24 bits)

Sub picture data max 32 streams Run length encoded(subtitles) Bit map



Some DVD-V features (2/2)• Seamless playback

Languageparental lockMulti-angle cameraStill pictureRegional coding (6 regions)

• System menuAudio stream selectionSubtitle selectionAngle selection

• EncryptionDecryption key hidden on the disc.



UDF & ISO9660

Part 3:Application Video

specification

Part 2:File system

Part 1:Physical

UDF

Read only

DVD-ROM DVD-Video

Audiospecification

Write-once

UDF | (UDF & ISO9660)

Rewritable

DVD-Audio DVD-R DVD-RAM

The DVD family of products



Recording on disk - principle• Products: CD-R, CD-RW, DVD+/-R(W) …

• CD principle: reflectivity of pits & lands are different.Pits and lands are used to store 0 and 1.

• CD-RW principle: reflectivity of the two phases of the recording material (amorphous, crystalline) are different.Controlling the phase allows storage of 0 or 1.

• To Amorphous state (low reflectivity):T above melting point (600°C) & fast cooling

• To Crystalline state (high reflectivity):T above 200°C for a sufficient time

• Recording: by the laser heating the recording layer

• Reading: by laser as for CD (-> compatibility)



Blu-Ray DVD

CD DVD Blu-Ray DVD Owner Philips & Sony many

companies Philips, Sony, Hitachi…

Data depth 1.2 mm 0.6 mm 0.1 mm Wavelength 780 nm 650 nm 405 nm NA 0.45 0.60 0.85 Spot dimension (relative)

1 1/2.6 1/13.2

Spot dimension (relative)

1 1/5.2

Capacity 650 MB (data) 4.7 GB (per side)

22.5 GB

Duration 70 min @1.4 Mb/sec

135 min @ 4.6 Mb/sec

150 min @ 20 Mb/sec

Focus VHS quality Standard TV HDTV



SACD• = Super-Audio-CD• = Response of Philips/Sony on the thread DVD-Audio

brings on the revenues from CD portfolio • Multi-layer hybrid scheme

– One layer for playback in CD player at standard quality– One layer for playback in SACD player at enhanced

quality (DVD-like, 4.38 Gbytes)

• Already on the market and in consumer homes (marketing argument)

• DSD technology (Direct Stream Digital)• delta sigma DAC to decode the 2.82 Mbps PDM

stream • Lossless compression, 5.1 multichannel, encrypted• 120 db (=20-bit), 100 kHz BW












• Conclusion



Adaptation to the DVB channel

• Channel coding : transforms the TS in an other sequence of bits containing the same information than the input stream but more robust against the imperfections of the transmission on the physical channel cost : a higher bit rate

• Modulation : transforms an input sequence to an analog waveform for transmission over the physical channel

Sequence of bits

(Encrypted) TS Channelcoding

Analog waveform

Modulation Physicalchannel



Channel coding (1/3)• Unlike source coding that removes redundancy,

channel coding adds redundancy in a structured way so that the decoder be able to detect and/or correct the errors introduced by the physical channel.

Channeldecoding

Source decoding

Sink

Source coding

Source Channelcoding

Quasi-error-freechannel (e.g.BER<1E-10)

Demodulation

Error-pronechannel (e.g.BER=1E-3)

Modulation

Physicalchannel



Channel coding (2/3)• Channel coding may include :

– Spectral modification of the signalfor adaptation to the channel (e.g. remove DC, spectrum shaping like uniform distribution in the frequency space ...)

– FEC : Forward Error CorrectionAddition of redundancy in order to allow error detection and/or correction (example : The total of bought articles is similar to a parity byte)

188 bytes

188 bytes

Allows correction of 8 erroneous bytes

After error correctionencoding

Original sequence

Message Parity bytes

16 bytes



Channel coding (3/3)– Interleaving

Time diversity in order to deal with error bursts.The successive bytes of information are dispersed in time on the transmission channel in such a way that an error burst does not affect neighbouring bytes. Interleaving is often combined with FEC so that error bursts could be corrected by the FEC.

Example :

AEIMBFJNCGKODHLPQ.....

AEIMBFJ&&&KODHLPQ.....

----> A burst of errors affects bytes belonging to different error correction blocks

Channel

Deinterleaving

AB&DEF&HIJKLM&OPQ.......|---|---|---|---|

Interleaving

ABCDEFGHIJKLMNOPQ.......

& : Erroneous byte| : Beginning of an error correcting block- : Element of an error correcting block



Modulation in DVB (1/3)• Different modulation techniques :

– Cable : QAM– Satellite : QPSK– Terrestrial : OFDM

• Why ?Modulation technique depends on :

– Physical characteristics of the channel– Compatibility constraints with actual analog

transmission



Modulation in DVB (2/3)

• Example : influence of SNR on modulation technique selected QPSK for satellite and QAM for cable

14

1E-7

4

1E-5

1E-6

6 8 10 12

QPSK

BER

0.001

1E-4

0.01

1.0

0.1

16 18

16QAM

20 22

32-QAM 64-QAM

24 26 28 SNR



Modulation in DVB (3/3)• Satellite

Bandwidth : generally 27-36 MHzSNR low : about 10 db (power transmitted by satellite)direct path

• CableBandwidth : 8 MHz (50Hz countries) - 6 MHz (60Hz countries)SNR strong (about 25 db)Echoes from impedance mismatch in the network

• Terrestrial Bandwidth : idem as cable Multipath interference, signal level variation, ...



From TS to the DVB channel

• Some blocks are identical for all standards (Cable, Satellite & Terrestrial)

• Inner & outer : terminology is derived from the view of the quasi-error-free channel composed of a transmitter and a receiver.

• Satellite & Terrestrial : More sensitive to error inner coder is added

QAMmodulation

QPSKmodulation

OFDMmodulation

DVB-C

DVB-S

DVB-T

Spectrumshaping

TS Outercoding

Inner inter-leaver

Innercoding

Inter-leaving












• Conclusion



Questions

?



APPENDICES



Agenda

Conditional access

• What is cryptography

• Symmetric & public-key cryptography

• Why cryptography for DVB ?

• Conditional access information in MPEG/DVB

• Conditional access mechanism

• Conditional access interfaces



• Why cryptography ?

– CONFIDENTIALITY - The message is not listened

– INTEGRITY - The message is not modified

– AUTHENTICITY - The message has been sent by Alice

– NON-REPUDIATION - Alice cannot falsely deny she has sent the message

What is cryptography (1/2)

ALICE

MESSAGE

BOB



What is cryptography (2/2)• Basic terminology

cryptographic algorithm or cipher

CiphertextEncryption

Plaintextor cleartext

Key Key

DecryptionOriginalplaintext



Symmetric cryptography Public-key cryptography Key1 = Key2 Key 1 Key 2

• Public-key cryptographyOne Public-key (known by everybody) : PKOne Private-key or Secret-key (kept secret) : SK

• C = EKey1(M) M = DKey2(C) = DKey2(EKey1(M)) In public-key cryptography, key1 may be PK or SK and key2 is the other key.

Symmetric & public-key cryptography(1)

Key 1

EncryptionPlaintextor cleartext

Cyphertext

Key 2

DecryptionOriginalplaintext




• Example of symmetric cryptography

– Key stream as long as message– Key stream = pseudo-random sequence (easy to

break)– Low security should be compensated by frequent

change of keys necessity of secure channel 2 channels : one for the message & one for the key

Pseudo-randomnumber generator Secure

channelKey

+Plaintext Ciphertext

Pseudo-randomnumber generator

Key

Original plaintext+




• Example of public-key cryptography

Alice encrypts messageusing Bob's public key

Public key of Bob

Encrypted message

Public Keys

Bob decrypts messageusing his secret key



Symmetric & public-key cryptography(4)• Symmetric cryptography example : DES

• Public-key cryptography example : RSA (1977)

• Symmetric versus public-key cryptography– Symmetric cryptography is faster (about 1000 times).

– Low security of symmetric cryptography (due to the necessity of key transport) is improved by a frequent change of the key.

– In Public-key cryptography the secret-key may be kept secret. It is never transported High security.

– Different usage : In DVB, symmetric key algorithm for encrypting data, public-key algorithm for key management (secure channel).

• Hybrid cryptosystemExample : DES for message and RSA for key encryption



Cryptography and DVB (1/2)• Cryptography prevents unauthorised receiver from

decoding the program.

• DVB compared with banking or military secret– high information rate

– low information value

– decryption must be cheap

• Cost of cracking the system should be higher than the benefits gained from the cracking

• Cryptography in DVB is a trade-off between cost/complexity versus piracy-proof.

• CA (Conditional Access) = very sensitive subject. Some service providers want their own CA system.



Cryptography and DVB (2/2)• MPEG does not specify a conditional access (CA) system but

defines a frame to support CA.

• DVB characterises some aspect left undefined by MPEG,It defines a CA interface.

• The broadcaster develops its CA system using a CA interface.

• DVB is based on– symmetric cryptography for audio-visual transmission

– frequent key change to increase security

– Public-key cryptography for key-exchange

• DVB relies on – stream of ECM’s (Entitlement Control Message)

– stream of EMM’s (Entitlement Management Message)



CA information in MPEG TS (1/2)

Multiplexing

VideoEncoder

ECM'sPES

STC

EMM'sPES

SYSTEMLAYER

COMPRESSIONLAYER

Transport Sublayer

PES Sublayer

ComputePCR

VideoPES

Packetising

PTS

Sampling

Video elementarystream (E.S.)

Audio elementarystream (E.S.)

Program SpecificInformation (PSI,SI)

MPEG2 Transport Stream

Packetising

PTS

AudioPES

TRANSPORTSTREAM

PACKETISEDELEMENTARYSTREAM

ELEMENTARYSTREAM

AudioEncoder

Uncoded video

Sampling

Uncoded audio



CA information in MPEG TS (2/2)

Table

MPEG TSMultiplex

Program Association Table (PAT) (PID=0)

Program Map Table (PMT) (PID=x)

Conditional Access Table (CAT) (PID=1)

Some possible tables

Program Number 1 2 ......

Stream-type Audio Video PCR's ECM's

Stream-PID aa bb cc dd

EMM's mm

PMT-PID x y......

EMM's

PCR

ECM's

Payload184 bytes

Header4 bytes

PID

Video

Audio



The CA mechanism : illustration

DecryptionEncrypted AV data Clear AV Data

SMARTCARD

Decryption

Decryption

ECM’s

(Program related)

EMM’s

(CA system related)

IK

EntitlementSK

Access control parameters

SK

CW’s

PDK1 PDK2PDK



The CA mechanism (1/2)

• AV streams are scrambled with Control Words (CW) using symmetric cryptography

• CW are encrypted using Service Keys (SK), are placed in ECM’s and are securely transmitted to the receiver

ACP = Access ControlParameters.

EncryptionCW (Control Word)

SK (Service Key)

Base Key

MPEG TS(clear) Scrambler

ECM's

Encryption EMM's

AV streams

Decryption

Decryption

ECM'sMPEG TS

EMM's

IK (Issuer Key)

AV streams

ACP

PDK (Programmer Distribution Key)

SK

EntitlementPDK,SK

Entitlement,PDK, SK

Descrambler

MPEG TS(clear)

CW



The CA mechanism (2/2)• SK are encrypted using public-key cryptography -

Keys are IK (unique key internal to the smartcard) or PDK (transmitted via EMM’s in order to define user’s group)

• ECM’s carries (informations related to a single program PID of ECM’s in PMT) – enciphered CW

– access parameters

• ECM’s are decoded to CW if the receiver contains the required entitlements

• EMM’s carries (information related to a conditional access system PID of EMM’s in CAT)– New entitlements, SK’s (Service Keys)

– Programmer distribution key



About DVB scrambling• Encryption occurs after compression (at the location

in the stream where the redundancy is at its lowest value) in order to have a robust encryption system.

• Encryption may occur at PES level or at TS level.

• DVB scrambling is transparent (a valid TS remains valid after scrambling) facilitates transport and manipulation.

• Synchronisation based on PCR’s constant time required for scrambling/descrambling.

• Security device should authenticate EMM’s origin.

• CA is only one aspects of cryptography usage in DVB. An other may be copy protection by (watermarking) and authentication (by signature).



Agenda

Some video format



Some video formats (1)

• Max. component video signal bandwidth: 6 MHz.

• CCIR601 (CCIR is now ITU-R): Video sampling frequency: 13.5 MHz for 525 & 625 line standards(Shannon requirement)

• Synchronous with line (& image) sampling frequencyFsampling= 864*Fh for 625 line system (50Hz countries) Fsampling= 858*Fh for 525 line system (60Hz countries)

• Why synchronous? Points at the same place

• RGB format



Some video formats (2)• YCbCr format

Cb = B-Y, Cr = R-YEye is more sensitive to luminance than to chrominance (lower resolution needed for chrominance)

R ed

B lueG reen

M atrixLP F

Y

C b

C r

M odula torS ubC

+C om posite

V ideo



Some video formats (3)• The 4:2:2 format

– Y sampling @ 13.5 MHz

– C sampling @ 6.75 MHz

– 8 bits per pixel

– 720 active points per line

– 576 lines active lines per image (2 fields) (625 lines)and 480 active lines (525 lines)

– Pixels are not square (e.g. for 480 lines, only 640 active points are needed - VGA format)

– Image size 720*576 or 720*480

• The 4:2:0 format– Vertical luminance resolution reduced by a factor 2

(average on two successive lines)



Some video formats (4)• SIF format (Source Intermediate Format)

Half the vertical & horizontal resolution of 4:2:0For 50Hz countries:– Luminance: 360*288

– Chrominance: 180*120

• CIF format (Common Intermediate Format)– Intermediate format used in videoconferencing

(communication between US & Europe)

– resolution: 360*288

– Sampling frequency: 30 Hz

• QCIF (Quarter CIF)– Half the vertical & horizontal resolution of CIF.

Audio & Video Compression and its Application in Consumer Products

Documents

Transcript of Audio & Video Compression and its Application in Consumer Products