digital-audio.org
Information centre for all about digital audioCompression
Computer data files can always be compressed, and since digital audio files are just data they can be compressed too. The reason for doing this is to save on storage space and to speed up transfer of files from one place to another. Compression works in various ways, but in principle it is done by replacing a block of data with a code that allows re-creation of the data. Compression programs will search the data for blocks that they can replace and if this can be done enough, then a considerable reduction in file size can be achieved. There are two distinct types of audio compression, Lossless and Lossy.
Lossless
Any audio file that is compressed can and will be expanded again to be played back, to reproduce the audio that was recorded. As the name suggests, 'Lossless' compression is a method by which nothing is lost in the compression and as such the original audio wave can be re-created on expansion exactly as it was before. This is similar to computer file compression because in that case, computers must have the file re-created exactly as it was or it will not work.
In terms of audio formats, examples of the Lossless compression formats are Windows Media Audio (WMA) Lossless, Apple lossless, and Free Lossless Audio Codec (FLAC). For more info on each of these formats please see the Glossary page.
Lossy compression
When audio files are compressed, they can still be quite large, and in some applications such as portable music players even smaller files are needed. This is where lossy compression comes in. This involves removal of some of the audio data as well as compression of what is left. That means that an audio wave can be recreated but it is not the same as the original.
There are various approaches, but in the main they all involve removal of unnecessary audio data. The theory used is that some sounds present in an audio wave are masked by other sounds, and the masked sounds can be removed as they are inaudible. The extent to which the masked sounds are removed depends on the compression level needed.
In comparison of Lossy formats the bit rate of the compressed file is usually quoted. This is because there is a direct correlation between the amount of data remaining and the rate at which bits of data are processed on replay. The bit rate can also affect the dynamic range and frequency response of the sound output depending on the format.
Examples of lossy compression formats are MP3, Apple's AAC, and Windows Media Audio. In general these formats can compress to produce a file who's bit rate varies between 64 kbps ( kilo bits per second ) and 320 kbps. Popular theory is that at least 192 - 256 kbps is needed to make the comparison between these files and the original all but negligible except to a very critical ear.
As a reference, Compact Discs on replay have a bit rate of 1411 kbps, and lossless compression formats have a replay bit rate of 600 - 900 kbps.