In general, people will code to the API they have. For graphics that mostly means GPU hardware and shaders. The old raster tricks still come out from time to time, and there are reasons to software render certain effects, but there are many new GPU-centric effects that are equally diverse and clever. Experienced devs will often use third-party engines as cross-platform deployment tools and then work around whatever the engine provides. But indies are relatively more likely to engage with the higher-level systems which can lead to a certain defining aesthetic(e.g. in the era of Flash games, the built-in filters, masks, etc. got a thorough workout, even though by the end, everyone making optimized games was caching their results to bitmaps).

With audio, on the other hand, everyone will just look for fmod or an equivalent. There's rarely much time budgeted for audio and there's much less shared experience in doing it ground-up.