2.1.17. Learn source range


The learn source range parameter allows you to enable or disable learning for source range.

While active, source range will be adjusted to follow the input loudness distribution rather than raw peaks. The learn algorithm uses LRA-style percentiles (10th/95th) from the input histogram and smooths updates using the current attack/release ballistics. This is useful for quickly setting the source range to a representative window of the input, then disabling learning to prevent the source range from changing.

Learn accounts for source distribution asymmetries, making it great for hitting a target range accurately.