Well, PS5 has an incredible audio processing chip that can actually use GPU geometry directly - and it does its own equivalent of audio raytracing to model accurate behaviour of audio inside whatever room you're in, based on geometry and material properties. It's extremely cool and works pretty well, but unless you're an audio geek you won't notice.
And actually, the "what has this hardware enabled" works kinda backwards. I work in video games and every single design that we come up with usually hits the wall of "oh that's cool on PS5/XSX, but how will we make this run on the Switch" - and the answer is we won't, so the feature gets scrapped. Increases NPC counts, better AI, multiplayer sessions with hundreds of people on the screen - these are arguably gameplay features that the current gen enables but because games are always made for the lowest common denominator they rarely get implemented. And yes, of course there are always tricks to do these things regardless, switch has some impressive games despite being positively ancient now. But I've been part of the discussion about so many gameplay systems that got scrapped not because they couldn't run on the PS5 but because they also had to perform well on lesser hardware.
AMD TrueAudio was doing the audio raycasting thing many years before the PS5 released [1]
> The TrueAudio Next SDK was released as open source through AMD's GPUOpen suite in August 2016...TrueAudio Next uses the GPU's ray-casting technology to do the audio computation
It's a rather niche feature, don't think too many games ever used it.
[1] https://en.wikipedia.org/wiki/AMD_TrueAudio