Audio compression is a technique for converting or encoding audio (sound) information so that a smaller amount of information elements or reduced bandwidth is required to represent, store or transfer audio signals. Audio compression coders and decoders (codecs) analyze digital audio signals to remove signal redundancies and sounds that cannot be heard by humans. Some of the basic coding processes include waveform coding, perceptual coding and voice coding.

Waveform Coding - Waveform coding consists of an analog to digital converter and data compression circuit that converts analog waveform signal into digital signals that represent the waveform shapes. Because waveform coders represent the shape of a signal, they can be used to represent complex audio signals (such as fax and computer modem signals).

Perceptual Coding - Perceptual coding analyzes and converts audio information into sounds that the ear can understand. For example, the ear cannot simultaneously hear loud sounds at one tone (frequency) and soft sounds at another tone (different frequency) and the human eye can sense differences in intensity (brightness) much more than it can sense changes in chroma (color). Using perceptual coding, is is not necessary to send signals that cannot be heard by a human.

Voice Coding - Voice coding is the analyzing voice sound information so that can be represented by a mathematical forma. Sounds that are not voice signals (such as music or fax) are filtered or ignored.

Frequency Masking - Frequency masking is the process of blocking, removing or ignoring specific frequency components of a signal (such as blocking high frequency signals from a telephone call).

Temporal Masking - Temporal masking is the process of blocking, removing or ignoring specific components of a signal that occur in a specific time period or time sequence (such as omitting a quiet sound that cannot be heard after a high volume sound has occurred).

This figure shows some of the frequency and temporal masking and techniques that can be used for audio compression. This diagram shows that the compression process may remove some audio information that the listener cannot hear. This example shows that the compression process may remove small sounds that occur directly after a louder sound (temporal masking). The compressor may also remove low amplitude signals that occur simultaneously with other frequencies (frequency masking). Compression systems may also ignore small changes in sound level.