Use as few multiband bands as the problem needs, and set crossovers deliberately

Every crossover in a multiband unit adds a filter that shifts phase, smears transients, and causes phase interactions between adjacent bands, so more bands than the problem requires means more coloration and side effects. The correct approach is to identify which frequency region has the problem and use only the bands needed to isolate it, with the fewest crossover points — often two bands (one crossover) when only the high or low end needs work. Default crossover frequencies in multiband plug-ins are rarely right for a specific problem; solo each band to verify the crossover is capturing the target region cleanly, and treat trusting the defaults as no better than using EQ presets. Also steer clear of presets generally, avoid automatic gain-compensation routines, and reassess attack/release per band since bass moves more slowly than treble. Multiband units can additionally be used in parallel for natural-sounding frequency enhancement.

Examples

A vocal that only needs de-essing above ~6 kHz is served by a two-band split; a four- or six-band unit here adds unnecessary crossover interactions. When only the high end is problematic, split at ~3 kHz and process only the high band rather than using all default bands.

Assessment

Explain why fewer multiband bands produce fewer artifacts. Describe the correct procedure for deciding how many bands to use and for setting the crossover frequencies, and why the defaults should not be trusted.