Independent Component Analysis
An unmixing method recovering independent source signals from linear mixtures, a stronger condition than PCA's decorrelation, able to separate blended signals.
Uncorrelated means no linear relationship; independent is stronger, ruling out nonlinear relationships too. ICA searches for the unmixing that makes recovered signals as independent as possible, using non-Gaussianity as a workable stand-in since independence itself is hard to optimize directly.
It needs at least as many recordings as sources, and it cannot recover original scale, sign, or ordering — only the shape of each source.