32/64-bit floating-point arithmetic in DAWs provides vast internal headroom but does not protect against plug-in overloading

A common misconception is that because 32- or 64-bit floating-point DAWs theoretically describe a dynamic range of ~1500 dB, any level is safe. In practice this is wrong for two reasons: (1) the rounding differences between floating-point calculations and the 24-bit audio portion can produce audible differences in summing when many high-level signals combine; (2) many plug-ins — including some from top manufacturers (Waves, Slate, Sonnox) — are designed to operate at a ‘normal’ level and degrade audibly when fed hot signals. Analogue-modelling plug-ins are especially susceptible, as accurate modelling of non-linear analogue behaviour at extreme levels is computationally expensive and often omitted.

Examples

Loading a soft-saturation plug-in on a channel peaking at -2 dBFS vs. -12 dBFS: the hotter signal produces unintended distortion that sounds digital and harsh, not the intended gentle warmth.

Assessment

A student argues that since their DAW uses 64-bit floating point, they can leave all faders at 0 dB and the built-in headroom will handle everything. Identify two specific scenarios where this causes audible problems.