Stereo Widener

The signal is split into mid and side, the sum of the two channels and their difference, the side is scaled by Width, and the two are recombined. Everything a stereo widener does follows from that: the mid is what both speakers agree on, the side is what they disagree on, and widening is simply turning the disagreement up.

This is also why the effect has a hard limit that no setting escapes. A mono source has no side signal at all, so there is nothing to scale and no width setting produces any change, feed it something genuinely stereo, either a stereo clip through the input panel or the chain's two oscillators panned apart, or this module will sit there doing nothing and be right to.

The browser preview does not build an explicit encode-scale-decode chain, because width control is already linear in L and R and collapses into a single 2x2 gain matrix. Substituting mid and side back into the output equations gives each channel a `same` coefficient on itself and a `cross` coefficient on the other, which is four GainNodes between a splitter and a merger and no summing stages beyond the merge itself.

Those coefficients are where the param range comes from. Width maps onto the side scaler as twice its value, which puts 0.5 at unity, the signal passes through exactly as recorded, 0 at full mono, and 1 at double side, where the image pushes out past the speakers and anything panned centre starts to feel hollow.