An Overview of Perceptual Audio Coding and MPEG AAC

1
An Overview of Perceptual Audio Coding and MPEG
AAC
2
Introduction

• Audio coding or audio compression algorithms are
  used to obtain compact digital representation of
  high-fidelity (wideband) audio signals for the
  purpose of efficient transmission or storage.
• The central objective in audio coding is to
  represent the signal with minimum number of bits
  while achieving transparent signal reproduction
  i.e. generating output audio that cannot
  distinguished from the original input even by a
  listener with Golden Ears
• The Motion Picture Experts Group (MPEG) audio
  compression algorithm is an International
  Organization for Standardization (ISO) standard
  for high- fidelity audio compression.

3
Continue

• MPEG audio compression standards are lossy audio
  coding standards. They try to compress audio by
  trying to reduce perceptual and statistical
  redundancies.
• The basic task of a perceptual audio coding
  system is to compress the digital audio data in a
  way that -
• - the compression is as high as possible, and
• - the reconstructed (decoded) audio sounds
  exactly (or as close as possible) to the
  original audio before compression

4
Audio Coding Techniques

• Parametric Coding
• Waveform Coding
• Time Domain
• PCM, DPCM, ADPCM etc.
• Frequency Domain
• Transform Coding, Subband Coding
• Hybrid Coding

5
Perceptual Audio Coding Basics

• Human hearing limited to values lower than 20kHz
  in most cases
• Human hearing is insensitive to quiet frequency
  components to sound accompanying other stronger
  frequency components
• Stereo audio streams contain largely redundant
  information
• MPEG audio compression takes advantage of these
  facts to reduce extent and detail of mostly
  inaudible frequency ranges

6
Generic Perceptual Audio Coding Architecture
7
Psychoacoustic Principles

• High-precision engineering models for
  high-fidelity audio currently do not exist. So,
  audio coding algorithms rely upon generalized
  receiver models to optimize coding efficiency.
• In the case of audio, the receiver is ultimately
  the human ear and sound perception is affected by
  its masking properties.
• Perceptual audio coders achieve compression by
  exploiting the fact that irrelevant signal
  information is not detectable by even a well
  trained or sensitive listener.

8

• Irrelevant signal information is identified
  during signal analysis by incorporating into the
  coder several psychoacoustic principles,
  including absolute hearing thresholds, critical
  band frequency analysis, simultaneous masking,
  the spread of masking along the basilar membrane,
  and temporal masking.
• By combining all these, a quantitative estimate
  of the fundamental limit of transparent audio
  signal compression i.e. Perceptual Entropy is
  determined for given audio frame.

9

• Perceptual entropy denotes minimum number of bits
  which should be allocated to a given audio frame
  to represent perceptually lossless audio.

10
Absolute Threshold of Hearing

• The absolute threshold of hearing characterizes
  the amount of energy needed in a pure tone such
  that it can be detected by a listener in a
  noiseless environment.
• It can be expressed with a non-linear
  function,
• Tq(f) 3.64(f/1000)-0.8 - 6.5e-0.6(f/1000-3.3)
  2
• 10-3(f/1000)4 (dB SPL)

11
(No Transcript)
12

• When applied to signal compression, it could be
  interpreted as a maximum allowable energy level
  for coding distortions introduced in the
  frequency domain.
• So using this information the noise levels during
  quantization are tried to fit below this
  threshold.
• Due to this quantization noise does not become
  audible.

13

• However
• The detection threshold for spectrally complex
  quantization noise is a modified version of the
  absolute threshold, with its shape determined by
  the stimuli present at any given time.
• Since stimuli are in general time-varying, the
  detection threshold is also a time-varying
  function of the input signal.
• A Spreading function helps to determine modified
  detection threshold of hearing in presence of
  stimuli in given audio frame.

14
(No Transcript)
15
Critical Bands

• Human ear can be viewed as a discrete set of band
  pass filters, which covers the entire 20kHz
  frequency range.
• The inner ear called as Cochlea contains
  frequency sensitive positions. Whenever any tone
  enters the cochlea it moves until it reaches the
  position where it resonates. (Works as spectrum
  analyzer)
• The critical bandwidth is a function of
  frequency that quantifies the cochlear filter
  pass bands. (unit Bark)

16

• As the center frequency goes on increasing, the
  bark-width also goes on increasing.
• Spectral analysis of audio content is performed
  using critical bands.
• Bark-width with center frequency f is gives as
  
• BWc(f) 25 75(1 1.4(f/100)2)0.69 Hz
• To convert frequency in Hz to Bark
• Z(f) 13 arctan(0.00076f) 3.5
  arctan(f/7500)2 (Bark)

17
Figure Idealized critical band filter bank
18
Masking

• Masking refers to a process where one sound is
  rendered inaudible because of the presence of
  another sound
• Simultaneous Masking (Frequency domain)
• Relative shapes of the masker and maskee
  magnitude spectra determine extent of masking
• Non-simultaneous Masking (Time domain)
• Phase relationships between masker and
  maskee determine masking outcome.

19

• Depending on the behavior of masker and maskee
  there are following cases
• Noise Masking Tone (NMT)
• Tone Masking Noise (TMN)
• Noise Masking Noise (NMN)

20
Noise Masking Tone Tone Masking NoiseWe
can see the asymmetry of masking power between
noise and tonal maskers. Significantly greater
masking power is associated with noise maskers
than with tonal masker.
21
Difference between SMR, NMR and SNR
22
Spread of Masking

• Masker centered within one critical band has some
  predictable effect on detection thresholds in
  other critical bands. This effect, also known as
  the spread of masking,
• It is often modeled in coding applications by an
  approximately triangular spreading function

23
Non-simultaneous Masking (Temporal Masking)
24
MPEG Audio Codec Family

• MPEG-1 (ISO/IEC 11172-3) Layer 2 (mp2)
• MPEG-1 Layer 3 (mp3)
• MPEG-2 (ISO/IEC 13818-3) AAC
• MPEG-4 (ISO/IEC 14496-3) AAC
• MPEG-4 HE AAC
• MPEG-4 HE AAV v2

25
MP3 Compression Flow Chart
26
Layer 3 uses a 2-stage filter, more frequency
resolution and improved Huffman Coding to the
basic perceptual coder principle
MDCT Filter bank
QMF Filter bank
27

• Bit rates available
• In MPEG-1 Layer 3 are 32, 40, 48, 56, 64, 80, 96,
  112, 128, 160, 192, 224, 256 and 320 kbit/s, and
  the available sampling frequencies are 32, 44.1
  and 48 kHz. 44.1 kHz is almost always used
  (coincides with the sampling rate of compact
  discs), and 128 kbit/s has become the de facto
  "good enough" standard, although 192 kbit/s is
  becoming increasingly popular over peer-to-peer
  file sharing networks.
• In MPEG-2 and the non-official MPEG-2.5 include
  some additional bit rates 8, 16, 24, 32, 40, 48,
  56, 64, 80, 96, 112, 128, 144, 160 kbit/s while
  providing lower sampling frequencies (8, 11.025,
  12, 16, 22.05 and 24 kHz)

28
Design limitations of MP3

• There are several limitations inherent to the
  MP3 format that cannot be overcome by using a
  better encoder. Newer audio compression formats
  such as Vorbis and AAC no longer have these
  limitations.
• In technical terms, MP3 is limited in the
  following ways
• Bitrate is limited to a maximum of 320 kbit/s
• Time resolution can be too low for highly
  transient signals, causing some smearing of
  percussive sounds
• Frequency resolution is limited by the small long
  block window size, decreasing coding efficiency
• No scale factor band for frequencies above
  15.5/15.8 kHz
• Joint stereo is done on a frame-to-frame basis
• Encoder/decoder overall delay is not defined,
  which means lack of official provision for
  gapless playback. However, some encoders such as
  LAME can attach additional metadata that will
  allow players that are aware of it to deliver
  gapless playback.
• Nevertheless, a well-tuned MP3 encoder can
  perform competitively even with these
  restrictions.

29
Advanced Audio Coding (AAC)

• It is a standardized, lossy digital audio
  compression scheme. It was developed with the
  cooperation and contributions of companies mainly
  including Dolby, Fraunhofer (FhG), ATT, Sony and
  Nokia, and was officially declared an
  international standard by the Moving Pictures
  Experts Group in April of 1997.
• Not backward compatible with other MPEG audio
  standards (like mp3)

30

• AAC was promoted as the successor to MP3 for
  audio coding at medium to high bitrates.
• AAC follows the same basic coding paradigm as
  Layer-3 (high frequency resolution filterbank,
  non-uniform quantization, Huffman coding,
  iteration loop structure using analysis
  by-synthesis), but improves on Layer-3 in a lot
  of details and uses new coding tools for improved
  quality at low bit-rates.
• Its popularity is currently maintained by it
  being the default iTunes codec, the media player
  which powers iPod, the most popular digital audio
  player on the market.
• Furthermore, the iTunes Music Store, whose sales
  account for 85 of the market for legal online
  downloads, sells AAC-encoded songs (encapsulated
  with FairPlay Digital Rights Management)

31
AAC's improvements over MP3

• Sample frequencies from 8 kHz to 96 kHz (official
  MP3 16 kHz to 48 kHz)
• Up to 48 channels
• Higher efficiency and simpler filterbank (hybrid
  ? pure MDCT)
• Higher coding efficiency for stationary signals
  (blocksize 576 ? 1024 samples)
• Higher coding efficiency for transient signals
  (blocksize 192 ? 128 samples)
• Can use Kaiser-Bessel derived window function to
  eliminate spectral leakage at the expense of
  widening the main lobe
• Much better handling of frequencies above 16 kHz
• More flexible joint stereo (separate for every
  scale band)

32

• Both the mid/side coding and the intensity coding
  are more flexible, allowing to apply them to
  reduce the bit-rate more frequently.
• An optional backward prediction, computed line by
  line, achieves better coding efficiency
  especially for very tone-like signals. This
  feature is only available within the rarely used
  main profile.
• Improved Huffman Coding In AAC, coding by
  quadruples of frequency lines applied more often.
  In addition, the assignment of Huffman code
  tables to coder partitions can be much more
  flexible.
• AAC and HE-AAC are far better than MP3 at very
  low bitrates, but at medium to higher bitrates
  the two formats are more comparable

33

• Modular encoding
• AAC takes a modular approach to encoding.
  Depending on the complexity of the bitstream to
  be encoded, the desired performance and the
  acceptable output, implementers may create
  profiles to define which of a specific set of
  tools they want use for a particular application.
  The standard offers four default profiles
• Low Complexity (LC) - the simplest and most
  widely used and supported
• Main Profile (MAIN) - like the LC profile, with
  the addition of backwards prediction
• Sample-Rate Scalable (SRS), a.k.a. Scalable
  Sample Rate (MPEG-4 AAC-SSR)
• Long Term Prediction (LTP) added in the MPEG-4
  standard - an improvement of the MAIN profile
  using a forward predictor with lower
  computational complexity.
• Depending on the AAC profile and the MP3 encoder,
  96 kbit/s AAC can give nearly the same or better
  perceptional quality as 128 kbit/s MP3

34
MPEG-2 AAC Flowchart
35
MPEG AAC Family
36
Extensions and Improvements

• Some extensions have been added to the
  original AAC standard
• MPEG-4 Scalable To Lossless (SLS)
• High Efficiency AAC (HE-AAC), a.k.a. aacPlus v1
  or AAC - the combination of SBR (Spectral Band
  Replication) and AAC used for low bitrates
• HE-AAC v.2, a.k.a. aacPlus v2 - the combination
  of Parametric Stereo (PS) and HE-AAC
• Perceptual Noise Substitution (PNS)
• Long Term Predictor (LTP) - added in MPEG-4 Part
  3.

37
MPEG AAC Performance

• MPEG AAC provides excellent audio quality.
  Reaching perceptually transparent quality at only
  64 kbit/s per channel, it fulfills the
  requirements for broadcast quality as defined by
  the European Broadcasting Union.
• With sampling rates ranging from 8kHz up to 96kHz
  and above, with bit rates up to 256 kbit/s, and
  with support for up to 48 channels, MPEG AAC is
  one of the most flexible audio codecs. Of course,
  the standard also supports mono, stereo, and all
  common multi-channel configurations (e. g. 5.1 or
  7.1).
• The low computational demands make AAC the ideal
  codec for any low bit rate high-quality audio
  application.

38
MPEG-HE AAC

• HE-AAC is the low bit rate codec in the AAC
  family and is a combination of the AAC LC
  (Advanced Audio Coding Low Complexity) audio
  coder and the SBR (Spectral Band Replication)
  bandwidth expansion tool.
• This combination achieves good stereo quality
  already at bit rates of 32 to 48 kbit/s. HE-AAC
  is also known as aacPlus and can be used in
  multi-channel operations.

39
MPEG-4 HE-AAC v2

• Combined with parametric stereo, the HE-AAC codec
  provides good audio quality starting at bit rates
  around 16 to 24 kbit/s for stereo content.
• HE-AAC v2 is also known as aacPlus v2.

40

• Rough work
• Explain basic psychoacoustic principles
  Absolute threshold of hearing, Critical bands,
  Phenomenon of masking Simultaneous, Masking
  asymmetry, Spread of masking, Non-simultaneous,
  Perceptual Entropy
• MPEG audio codec family mp3, mp2 AAC, mp4 AAC,
  advanced AAC plus version 1, advanced AAC plus
  version 2
• (mention features present/absent in each)

41

• Limitations of mp3
• What is different in AAC ?
• Features in AAC
• Explain each feature in detail (mp2, mp4)