Back in the late 90′s and early 2000′s there was a lot of movement in the sound card industry – running alongside the GPU race. However, for various reasons the advancements in sound began to slow down. One of the major reasons behind this was actually Microsoft’s Windows Vista. It moved most of the audio stack to software (wiki link). This was a significant shift from previously, as they had been mostly handled through the sound cards drivers. This meant that a lot of the processing could be shifted on the sound card. In short, when 3d sound was needed to be processed the sound card was told what to do and it pretty much went ahead and did its thing. Vista simplified this, and also many motherboards were fitted with a Realtek AC97 audio compatible sound system and therefore, much of the progress was lost.
Microsoft’s Xbox One also has a custom built audio processor – but that’s being used for many things, including kinect. For AMD’s Radeon graphic card they’re using the Tensillica audio DSP – which is a somewhat off the shelf component. It’s somewhat of a mixture of a fixed function and a fully programmable processor. This means that power costs along with the size of the thing are minimal and yet is pretty damn flexible. From what we understand much of the hardware is being programmable in versions of the C language. We don’t have huge amounts of info on their inner workings yet, we can possibly guess some type of VLIW (Very Long Instruction Word – wiki link) design.