5. Structured audio rendering and perceptual optimizations

As we saw in the previous paragraph, rendering a 3D sound source involves a large number of signal processing operations. Even in the case of simplified models, performing all these processing operations for a large number of sound sources remains too computationally expensive for most applications. What's more, hardware rendering solutions [134] (see [TE 5 912] ) only support a limited number of simultaneous sound sources, also known as "channels" (from 16 to 64 for current cards, 128 for the next generation). Yet it's clear that a large number of sound sources may be necessary to...