You may use the make-up gain and automatic source normalization to adjust the volume of each of your audio sources. These volume controls, as opposed to other volume controls in Wwise, are transparent from all logical behavior based on volumes. For example, they have no effect on whether a voice is considered below threshold (virtual), and they are ignored when evaluating HDR attenuation. They don't appear in the Voice Monitor either.
The total value of these source-specific gains is displayed in the column Normalization / Make-Up Gain in the Voices tab of the Wwise Profiler.
The loudness data collected in the analysis phase of your original files may be used by Wwise to automatically normalize your assets at run-time based on their estimated loudness. Loudness computation loosely follows the ITU-R BS 1770 recommendation for loudness measurement: K-weighting filtering and absolute and relative gating at -70 dB and -10 dB respectively, on 400 ms windows with 75% overlap.
Source normalization is non-destructive: the sound's analyzed loudness value is stored separately by Wwise, and a proper normalization gain is applied at run-time. This gain is a function of the loudness: gain = -loudness - 23 [dB]. For example, if a sound's analyzed loudness is -37 dB and source normalization is enabled, Wwise will apply a normalization gain of +14 dB (+37 - 23) at run-time. If a sound's analyzed loudness is -12 dB, the normalization gain will be -11 dB (+12 - 23). In other words, soft sounds are boosted and loud sounds are attenuated. If you are unhappy with the normalization gain that is calculated based on loudness, you may tweak it further using the make-up gain.