Reproduction:
- Unmix Component (from a single source, this is just to create layers)
- Group transient and noise layer together
- make a time selection with the group as the selected target layer
- Apply -50 dB gain reduction to the selection
Results: one of the layers will accumulate gain, the other will attenuate some of the selection, but not all of it (almost like the erase tool with amplitude limits).
Expected results: Each layer comprised of the group will have the module applied to them individually (in this example, each layer has -50dB gain applied to them independent of their collective sum).
Am I misunderstanding this workflow? I had worked like this at one point