Why A speech processor raises average talk power by compressing or clipping the audio, but overdoing it hurts all three ways at once. Too much compression flattens and clips the waveform, which is audible distortion; that clipping generates harmonics and intermodulation products in the audio that translate into splatter and wide, adjacent-channel signals; and because compression raises the gain during pauses, it also pulls up shack noise, fans, and background sounds that would otherwise be inaudible. The goal is a few dB of compression with ALC just barely moving, not maximum processing.
Watch out Each single-effect choice is true but incomplete, so picking only distorted speech or only intermodulation misses the other real consequences of over-processing.
Over-processing: distorted, splattery, and noisy. If one processor symptom is listed, expect all of them to be.