Because software audio sucks major balls. It's bland, everything sounds the same "plasticky" with fake material effects attached to them and has appalling 3D space positioning where you can hardly tell where enemies really are.
I'm just playing System Shock 2, a GOG version that supports OpenAL and 48 channels with EAX 2.0 (this game supported EAX 2.0 even back then, just through DS3D and with only 16 channels) and the audio segment of it blows any modern game out of the sky. And it's a freaking title from 1999. When you walk over metal surfaces you actually feel like you're walking on metal. Hit a metal wall with a wrench and you can just sense the vibrations of the wall. Hitting an indestructible glass window gives the same vibe. Dropping rifle on top of a rifle with metal colliding noise or dropping a bottle of orange juice, the noise of echoing machines rumbling in a room or a hall, the echoing distant noises of hybrids who mumble to themselves etc. It's just not comparable.
Or a simple comparison of Need for Speed 3 Hot Pursuit where you race through a tunnel with a car using EAX. It sounded so good you just can't describe it. Not a single NFS game after Porsche 2000 (basically when they scraped EAX support) is capable of capturing that because reverberation plain and simply sucks.
Of course you don't see any difference in performance if everyone use shitty software audio with shitty cheap effects just so they don't cause any drag on the CPU. EAX 5.0 supports real time audio reflection and occlusion, basically audio ray tracing with multiple effects attached to each reflection and with soft transitions between environment effects. Drag wouldn't be as big as it used to be 20 years ago, but it would still be a problem on weak CPU's. Until you hear that in motion, I'm sure you don't miss real proper audio. Growing up with Aureal Vortex powered soundcard and the 6th generation of Sound Blaster, I'm someone who greatly appreciates high quality in-game audio. Just making loud explosions and talking about soundcards that have SNR of 3 billion decibels means nothing to me. That's not how you measure audio quality in games. Making sound that gives you the immersion of actually standing there, however does. It has to feel organic with the game, not disconnected from it like all these shitty software audio engines used by 99.9% of games released these days...
