I understand that you've limited the far distance of avatar audio sources, forced them to an inverse square falloff, and forced near/volumetric radius to zero to ensure that they actually get quieter as you move away. This is all well and good and should hopefully reduce problems with very loud avatars - however, when designing an avatar that is intended to not cause a disturbance to others nearby it presents a conundrum - if you set the maximum value for far radius, you'll potentially disturb people far away, but if you set it to be short, then falloff is unnaturally sharp for people nearby. This may also cause issues for audio sources that are specifically designed to only be audible very close to their origin (heartbeat effects, for example), as now the audio volume will change quite rapidly with any small change in distance from the source.
To help address these issues, please provide a way that avatar audio sources can use a gradual falloff nearby, then transition to a falloff curve sharper than normal at a configurable distance. This could be designed to be no louder than the maximum distance falloff curve used in the beta currently.