Abstract: A method for computing the “clippings” of an audio signal in the digital domain to prevent aliasing is disclosed. An offset signal is found by subtracting a threshold from samples of the audio signal. These are multiplied by a pulse after the pulse is lined up in time with the crossing of the audio signal at the threshold. The pulse is the sum of two step functions, both of which are bandwidth-limited.

Abstract: A digital clipper is highly oversampled to decrease aliasing and increase accuracy. The difference between the clipper's input and output is then downsampled and added to the delayed, unclipped signal at 1× sample rate to achieve clipping. Filters operating at 1× can be placed in series with the downsampled differentially-clipped signal to achieve overshoot compensation, bandlimiting of the clipped signal, and other goals.

Abstract: An audio peak limiter in which the lower sideband of the modulation distortion spectrum is canceled. The peak limiter is realized by multiplying the input signal by a control signal. The Hilbert transform of the control signal is multiplied by the Hilbert transform of the input signal and the outputs of the two multipliers are added. This cancels the lower sideband of the modulation process. Various filters can be added to eliminate aliasing in digital realizations, to band-limit the output signal, and to control peak overshoot.

Abstract: An audio peak limiter suitable for use between a compressor and pre-emphasis circuit. The limiter uses a center clipper to develop a control signal for a 1/x control law VCA. The center clipping, after rectification, is sampled-and-held for 4 milliseconds by a retriggerable sample-and-hold circuit then low-pass filtered. The compressed audio is delayed for 4 milliseconds before being coupled to the VCA. The VCA acts as a modulator to provide control of its peak output level while simultaneously limiting the bandwidth of the modulation sidebands caused by the modulation process, so that these sidebands are better masked psychoacoustically than sidebands that would be caused by simple clipping of the compressed audio.