The application of FILE COMPRESSION algorithms to digital sound files to reduce their required storage space and hence their download time over low-bandwidth communication links such as the Internet. Lossless audio compression can be achieved using exactly the same algorithms as for any other kind of digital data. However, as with visual images, compression is most effective when some quality is sacrificed (so-called LOSSY compression methods), and the most widely used compressed formats such as MP3 and REAL AUDIO use such methods to achieve compression ratios on the order of 60:1 to 80:1.
The easiest way to reduce the information content in a sound file is to lower the SAMPLING RATE and WORD SIZE below those required for CD-QUALITY music (44 kHz and 16- bit samples). To increase compression further, a variety of sophisticated algorithms have been invented, based on psycho-acoustical research into human sound perception. ‘Perceptual noise shaping’ devotes more bits to encoding those frequency components to which the ear is most sensitive, while other techniques discard altogether those sound frequencies to which the human ear is least sensitive: for example the very highest and lowest frequencies, and frequency components that are momentarily drowned out (so called ‘psycho-acoustical masking’) by a louder sound at a more sensitive frequency. Those low frequencies that are retained may be combined into a single monoaural sound instead of stereo, since the ear lacks directionality at low frequencies.