Editing Dbx (noise reduction) (section)

===Companding===
[[File:Dbx_compression_overview.svg|thumb|dbx companding compresses the original source (left) into a version with less dynamic range (middle), and then re-expands it (right). The tape hiss (pink) is also expanded by this process, but is overwhelmed by the now-expanded original signal.]]
dbx Type I and Type II are types of "companding noise reduction". These systems work by first compressing the dynamic range of the signal into a range that can be safely recorded on the tape. This type of compression, [[dynamic range compression]], mutes down loud sounds and amplifies soft ones, making the volume of the recording much more even. On playback, the dynamic range is expanded by the same amount, causing the low-volume sounds to become low-volume again and vice versa. The combination of compression and re-expansion gives rise to the name [[companding]]. Companding is useful even outside the field of noise reduction; a cassette might have 40 decibels of dynamic range before the media saturates, while the original signal might use 70 for, say, a live recording of a concert. In this case, companding at 2-to-1 will result in a signal with 35 decibels of range, which can be recorded without clipping.<ref name="Nave_2001_4"/>

The reason this technique works for noise reduction is that the tape hiss manifests itself as a constant low-volume signal. When the signal is recorded in its original form, without compression, the amount of hiss may be the same volume as softer sounds, masking them entirely. However, when the signal is compressed before recording, those soft sounds are recorded at a louder volume, so now even the soft sounds are louder than the noise. This improves the [[signal-to-noise ratio]].<ref name="Nave_2001_4"/>

When the signal is re-expanded, the tape hiss is expanded along with it, making it louder as well. However, the ''ratio'' of the signal to noise remains (close to) constant through this process, so the resulting output retains this higher signal-to-noise ratio. Ultimately, it means that while tape hiss does get louder during "soft" portions of the recording, the recording itself is (hopefully) always greater in volume and renders the hiss much less noticeable.<ref name="Nave_2001_4"/>