Thread: Soundcard FAQ
View Single Post
  #6 (permalink)  
Old 9th December, 2002, 04:16 PM
Aedan Aedan is offline
Chief Systems Administrator
Join Date: September 2001
Location: Europe
Posts: 13,075

Encoding systems

What is an encoding system?
In terms of audio, an encoding system is a method for either ensuring the audio survives transmission, or a method for reducing the space the audio takes. On occasions, an encoding system can do both!

What types of compression are there?
There are two main types of compression that can be used. One form is lossless compression, and the other form is lossy compression.

Lossless compression manages to keep all the information that was originally recorded, but squeezes it into a smaller space. When at the other end, the squeezed version is expanded back into it's original form, and is exactly the same as the original recording.

Lossy compression uses various techniques to analyse the signal, and work out which bits you won't notice go missing. This throwing away of bits helps keep the information small, and can be used to help a lossless compression to it's job better. The most popular lossy compression techniques use a model of the human hearing to work out which parts can be thrown out. Lossy compression is often based on psychoacoustic compression.

Lossy Compression Techniques

Temporal masking
Temporal masking is an effect of the way that humans hear. If you play a single note at one frequency, it takes time before the ear can hear a another quieter note at a close frequency. This effect is the temporal masking.

Frequency masking
Frequency masking is an effect of the way humans hear. If you play two notes that are close, and one is louder than the other, the ear may only hear the louder of the two notes. This is frequency masking.

The Discrete Cosine Transform is a method of converting temporal information into the frequency domain. A spectrum analyser is a device that can use a DCT to convert the sound over time into a set of frequency bands.

Sony's Adaptive TRansform Acoustic Coding was designed for use with their MiniDisc players. As the MiniDisc only held about 80Mb of data, Sony needed some way of compressing 650Mb of data into that 80Mb space. ATRAC was the result of Sony's work.

Officially known as MPEG1 layer 1. Layer 1 uses a basic DCT filter with one frame, and only uses frequency masking.

Officially known as MPEG1 layer 2, layer 2 builds on the foundations of layer 1. Layer 2 uses three frames, and starts to model temporal masking.

Officially known as MPEG1 Layer 3, layer 3 builds on layer 2. Layer 3 improves the filtering, includes temporal masking, can take stereo redundancy into account, and uses a Huffman coder.

The WMA format is a Microsoft propriatory format. It includes digital rights management, and is probably based around similar techniques to MP3.

Real Audio
The RealAudio format is a Real Networks propriatory format. It includes digital rights management, and is also probably based around similar techniques to MP3.
Any views, thoughts and opinions are entirely my own. They don't necessarily represent those of my employer (BlackBerry).