One ubiquitous type of audio encoding is pulse-code modulation (PCM), which is simply a linear (proportional) representation of the sound waveform. With PCM, the number stored in each sample is proportional to the instantaneous amplitude of the sound pressure at that point in time. The numbers are frequently signed or unsigned integers. Besides PCM, other encodings include mu-law and a-law, which are nonlinear mappings of the sound amplitude that are often used for recording speech.

You can use a predefined encoding by referring to one of the static objects created by this class, such as PCM_SIGNED or PCM_UNSIGNED. Service providers can create new encodings, such as compressed audio formats or floating-point PCM samples, and make these available through the AudioSystem class.

The Encoding class is static, so that all AudioFormat objects that have the same encoding will refer to the same object (rather than different instances of the same class). This allows matches to be made by checking that two format's encodings are equal.