[F4CS] File formats and Compression

• show understanding that sound (music), pictures, video, text and numbers are stored in different formats
• show understanding of the concept of Musical Instrument Digital Interface (MIDI) files, JPEG files, MP3 and MP4 files
• show understanding of the principles of data compression (lossless and lossy) applied to music/video, photos and text files

Objectives

Data compression

• Reduce the size of data, using compression algorithm, during transmission or store in file
• Advantages:
  • Faster transmission (less bandwidth required)
  • Save storage space
• Disadvantages:
  • Slower access time - data must be decompressed before use
  • More memory and processing time is needed

Lossless compression

• The compressed data can be recovered (decompressed) without loss of data
  • i.e. the original data before compression can be 100% retrieved after the data is compressed and decompressed
• Common application:
  • File compression (.zip, .rar)
  • Text file (.txt)
  • Transmitting data through Internet (e.g. HTTP data are often compressed nowadays)
• Lossless compression are not very effective in multimedia files

How Lossless works?

• One lossless compression algorithm is called RLE (Run-length-encoding)
• e.g. consider the following string of pixels:
  • BBBB BBBB WWWW BBWW - 16 Bytes
• RLE will only record the pattern and how many repetitions such that the compressed string became:
  • B8 W4 B2 W2 - 8 Bytes
• Other lossless compression algorithm may work on longer repeating patterns (e.g. the word algorithm appears 3 times on this page, we can give that word an index (number) and replace all occurrences with that number) 

Lossy compression

• During compression, some data is removed permanently (cannot retrieve after decompressed)
• Lossy compression works well in multimedia files, 
  • e.g. a 20MB picture can usually reduce to 1MB without sacrificing a lot detail using JPEG
• Common lossy compression files:
  • JPEG (or jpg)
    • NOTE: PNG is lossless
  • MP4
  • MP3

How Lossy compression works?

Original image (pixel value)

Lossy compressed

• Consider the above as a portion of an image, showing the value of individual pixel
• Lossy compression algorithm will group pixels with similar color (value), and assign them with same color (usually average)
• The image now consisted of many repeated pattern, so method such as RLE can be applied to reduce the size

• The rightmost image is highest compressed so those color patch (neighbor considered as "similar" are grouped) are very obvious, while the middle one is almost not noticeable compare to original
• Note about the efficiency of lossy compression, the middle image is almost identical to the original but size is 85% smaller

Audio file formats

• Uncompressed sound waves are stored as .wav files, which stores the amplitude of the sound waves
• .mp3 is a type of lossy compression works on audio data
• e.g. a 3-minute CD quality audio file is about 30 MB in size, while compressed in mp3 is about 2-3 MB
• MP4 is a file format for audio and video, also a standard format for internet video nowadays

MIDI

• Musical Instrument Digital Interface
• A file format (.mid) and also refers to the protocol of electronic instruments
• When music is stored in MIDI, it is not recording the sound wave, but the followings:
  • Types of instrument (e.g. Grand Piano)
  • Note played
  • Loudness
  • Duration etc.
• MIDI can only store music (without vocal) 
• Very small file size