Parameter Automation in a Dynamic Range Compressor

Dynamic range compression is a nonlinear, time dependent audio effect. As such, preferred parameter settings are difficult to achieve even when there is advance knowledge of the input signal and the desired perceptual characteristics of the output. We introduce an automated approach to dynamic range compression where the parameters are configured automatically based on real-time, side-chain feature extraction from the input signal. Parameters are all dynamically varied depending on extracted features, leaving only the threshold as a user controlled parameter to set the preferred amount of compression. We analyze a series of automation techniques, including comparison of methods based on different signal characteristics. Subjective evaluation was performed with amateur and professional sound engineers, which established preference for dynamic range compressor parameters when applied to musical signals, and allowed us to compare performance of our various approaches against manual parameter settings.