Abstract: A noise reduction system compresses an input signal at a constant ratio in response to a control signal derived from the compressed signal. Added to the derived control signal is a counterfeit control signal which causes the compressor to mistrack when the compressed signal falls below a preselected threshold level. Thus, low level compressor output signals are increasingly deemphasized. A constant counterfeit control signal produces a logarithmic decrease in compressor output signal levels below the threshold point. The resulting compressed signal is used in two ways to enhance system noise reduction. First, the compressed signal is expanded at a constant ratio over all signal levels so that low level mistracked signals are further deemphasized and the system noise floor is lowered.

Abstract: A flat frequency response compansion noise reduction system employs both input signal and control signal preemphasis and deemphasis in its encoder and decoder networks respectively. The linear range of the frequency response curve of the input signal deemphasis is selected to approximate the frequency range within the audio spectrum at which the human ear is most sensitive, typically having a high frequency end of 3.617 kHz. The frequency response curve of the input signal preemphasis is a mirror image of the frequency response curve of the input signal deemphasis. The linear range of the frequency response curve of the control signal preemphasis is selected to provide an encoder output signal having a substantially flat response. The frequency response curve of the control signal deemphasis is a mirror image of the frequency response curve of the control signal preemphasis. The amount of preemphasis is maximized to the limits of the recording medium, typically ranging from 0 db to 20 db.