Audio compression

We generally differentiate between lossless and lossy audio compression. With lossless compression, the output signal can be restored bit-identical from the compressed audio data. These methods typically achieve compression rates of 25 to 70 percent for typical audio contents such as music. Lossy audio compression on the other hand results in irrevocable loss but also yields higher compression rates. A compressed audio signal can never be restored to the exact original form. These processes often use so-called psychoacoustic models. This takes the capabilities of the human ear and personal audio perception into account. For example, due to shortcomings in human hearing, certain parts of audio signals can simply be omitted without ever being detected.

Telephony via Voice over IP (VoIP) and cloud phone systems also use audio compression processes. The voice quality of these compresses can be defined via MOS (Mean Opinion Score). The higher the MOS, the better the perceived voice quality. The indicator for the minimum bandwidth required for a phone call to be of adequate quality is often defined as ten kilobits per second. However, many codecs are used which require significantly higher bandwidths. Other parameters such as overhead or encryption also need to be considered. The codec G.711 is considered as somewhat of standard codec in digital telephony. A Voice over IP environment often uses the G.729a codec.