Audio processing

Audio processing changes a digital sound signal before it reaches an audio device. BetterDisplay's Software Audio can change level, frequency response, stereo channels and dynamics along one processing path.

An equalizer, or EQ, changes the level of selected frequencies. A low or high shelf affects the range below or above its chosen frequency. A peaking band affects a region around its center. A high-pass filter removes low frequencies, a low-pass removes high frequencies, and a notch reduces a narrow region. Q describes selectivity, with higher values producing a narrower region. The Nyquist limit is half the sample rate, above which a digital signal cannot represent distinct frequencies. The W3C Audio EQ Cookbook describes the causal filter equations used here.

Audio headroom is the margin available before digital clipping. Positive EQ or stereo widening can increase peaks, so automatic attenuation can reserve headroom. dB is a relative level measure. dBFS means decibels relative to digital full scale, with 0 dBFS marking the maximum representable sample level. RMS describes average signal energy. These digital quantities do not measure acoustic sound pressure at your listening position.

A compressor reduces gain above a threshold. Ratio determines how strongly it compresses, attack and release govern timing, knee softens the threshold transition, and makeup gain raises the processed level. A noise gate attenuates quiet signals below a threshold. Hold delays closing and hysteresis uses different opening and closing conditions to reduce rapid switching.

Temporary muting during startup, wake or configuration restoration is a separate coordination step. Ordinary volume, mute and tuning adjustments do not add a temporary software mute. During restoration, BetterDisplay briefly allows the processor to render digital silence before sending hardware settings. This transition is limited to half a second after processing becomes active. It resumes the selected software volume and mute settings when commands are sent, without waiting for hardware confirmation. A missing processing acknowledgement or a hardware-confirmation timeout produces a warning while requests and playback continue. This is not an acoustic level measurement or a guarantee of protection after processing ends.

These causal processors keep filter and envelope history without looking ahead or adding a playback queue. They still have response time and frequency-dependent delay. Compression without lookahead cannot guarantee catching every transient. A gate can alter the beginning or end of a sound. Stereo-linked detection applies the same gain change to both channels.