Methods and systems are provided for generating correction estimates. Training signals may be injected into one or more particular spectral regions, and one or more correction estimation parameters may be determined based on the injecting of the training signals, where the one or more correction estimation parameters reduce distortion in at least the one or more particular spectral regions. The particular spectral regions may comprise originally-unoccupied spectral regions. The one or more correction estimation parameters may be applied during correcting of digital signals generated based on processing of received analog signals. The training signals may be generated, such as based on one or more pre-defined characteristics. The one or more correction estimation parameters may then be determined based on the one or more pre-defined characteristics of the training signals and/or changes thereto.