MPEG3 For Audio

1
MPEG-3 For Audio

• Presented by
• Chun Lui
• Sunjeev Sikand

2
History of MP3

• In 1987, the Fraunhofer IIS started to work on
  perceptual audio coding in the framework of the
  EUREKA project EU147, Digital Audio Broadcasting
  (DAB).
• In a joint cooperation with the University of
  Erlangen (Prof. Dieter Seitzer), the Fraunhofer
  IIS finally devised a very powerful algorithm
  that is standardized as ISO-MPEG Audio Layer-3
  (IS 11172-3 and IS 13818-3).
• MPEG-3 Moving Picture Experts Group Audio Layer
  3

3
Background How CD stores music

• Music is sampled at 44,100 times/second
• Each sample is 2 bytes
• Sample is taken separately for left and right
  speakers
• 44,100 samples/sec 16 bits/sample 2 channels
  1,411,200 b/sec
• 3-minute song - 1,411,200 b/sec 180 sec
  31,752,000 bytes
• Yikes!

4
The 3 MPEG layers
5
MP3 Performance
6
How Does It Work?

• Every MP3 encoder uses two approaches
• First it compresses the analog audio stream with
  the help of perceptual noise shaping.
• Second, the compressed and frequency cleansed
  data is shrunk again using Huffmann encoding.

7
Perceptual Noise Shaping

• Much like YUV encoding and other compression
  scheme, it takes advantage of Psychoacoustics
• Human ear cannot distinguish the difference
  between two similar frequencies
• There are sounds that the human ear cannot hear
• There are sounds that the human ear hears much
  better than others
• If there are two sounds playing simultaneously,
  higher volume mask lower volume

8
Perceptual Noise Shaping Example

• If there is a louder sound in one band, dont
  really need to encode all the other bands

9
Encoding a Wave File in MP3
10
Analytical Filter

• The audio signal passes through a filter bank
  which divides the audio signal in 576 areas
  (sub-bands). This requires very complex filters.
  Here, MP3 encoders work with the well-known
  Discrete Cosine Transformation.
• In real time it calculates unnecessary
  frequencies and eliminates them iteratively
  (repeatedly) until the best possible result is
  achieved.

11
Masking Threshold Evaluation

• At the same time, the audio signal passes through
  the psychoacoustic model. For every sub-band of
  the entire signal spectrum, the masking threshold
  is determined using the Discrete Fourier
  Transformation.
• Joint stereo coding can then be done to take
  exploit the fact that both channels of a stereo
  channel pair contain by far the same information.
  These stereophonic irrelevancies and redundancies
  are exploited to reduce the total bitrate.

12
Quanitzation and Encoding

• When quantizing the sample another starting point
  for data reduction arises. Every sample is made
  up of 16 bits, but not all 16 are necessarily
  needed in order to represent the sound. As such,
  the leading nulls of a 16-bit sample may be left
  out
• At the same time, individual samples are analyzed
  and compressed again using Huffman encoding. This
  produces a further reduction of data of about 20
  percent.

13
Bitstream creation

• Now all the data has been gathered.
• Everything is recorded and digitalized.
• Finally, the encoder forms the bit stream, which
  ultimately represents the MP3 file the
  compressed data is compiled into so-called
  frames.
• For MP3s, there are 1152 scanning values per
  frame (32 Sub-bands 36 Samples). Every frame
  consists of a header, a sum test check, the audio
  data and sometimes a bit-reservoir.

14
MP3 Bit Rate

• Measure at bits per second
• Generally speaking the higher the bit rate of a
  MP3 file the higher quality it is
• There are 3 types of bit rate format
• Constant Bit Rate (CBR)
• Variable Bit Rate (VBR)
• Average Bit Rate (ABR)

15
Constant Bit Rate

• Same bit rate for the entire file
• Stream an audio more efficiently
• Quality of the encoded content is not constant
• Because some content is harder to compress than
  others
• Different song encoded with same bit rate may
  result in different quality

16
Variable Bit Rate

• Try to achieve best quality
• Use different bit rate for each frame of the MP3
  file
• Use a higher bit rate if needed
• Better quality
• Result with a larger file size
• Many MP3 players do not support VBR

17
Average Bit Rate

• A little like VBR
• Use the average bit rate of all the bit rates VBR
  would use
• Dynamically change the for better quality
• The end file size is known
• Global quality is slightly lower than VBR

18
MP3 Competition

• Window Media Format - CD quality at half the bit
  rate
• Real Audio - Designed for lower bit-rate
• Liquid Audio - implements more security so less
  appealing to users

19
Acknowledgements

• Fraunhofer IIS
• http//www.iis.fraunhofer.de/amm/techinf/layer3/i
  ndex.html
• Intel http//www.intel.com/english/home/maximize/
  article/mp3/how/index.htm
• Koning, Verhelst. ON PSYCHOACOUSTIC NOISE SHAPING
  FOR AUDIO REQUANTIZATION. Vrije Universiteit
  Brussel.
• msdn.microsoft.com/library