MrAudioSoundImages's video: Noise Reduction Software

Analog noise reduction devices like dbx and Dolby achieve noise reduction by companding, meaning that the audio is dynamically compressed when recorded and then dynamically expanded on playback. Digital noise reduction software, however, works more like Burwen and other analog noise reduction devices that gate the signal on playback. Some of the earliest available noise reduction software includes DINR (Digidesign's Intelligent Noise Reduction) in 1994 and Sonic Solutions No Noise introduced around 1996. More recently companies like Izotope, Waves, and others have introduced plug-ins to minimize noise. In addition to software, companies like CEDAR have hardware processors (as well as plug-ins) that are effective at reducing noise and hum, surface pops and clicks, and even minimizing distortion. After the cold war ended in 1991 former KGB agents who had developed sophisticated software for enhancing intercepted telephone conversations went into business selling their software to the private sector under the brand name of Speech Technology Center, and their software is still widely used for forensic applications. In any case, digital noise reduction software works by first establishing a threshold level, usually by capturing a small sample of the noise. Then, by using a series of up to 200 (or more) narrow band dynamic filters, each targeted to a specific frequency range, the audio signal is gated (turned off), whenever the level falls below the threshold at that particular frequency. Care needs to be taken to balance the noise reduction against the artifacts that can be produced in the process, since noise reduction always works the best when you need it the least.