Other answers address fan construction and fan control. I think it’s also important to consider the overall system design and target audience.
Early PCs, in fact PCs until the late 486 and Pentium era (early to mid 90s) typically had a single fan, which had to take care of all the ventilation requirements of the system, not only as built, but as the user might end up equipping it. This single fan was included in the PSU, which restricted its size, and required a higher speed to produce a given airflow than is possible with today’s 12cm-fan-equipped PSUs. Furthermore, while PCs themselves had smaller thermal envelopes than today’s PCs, the PSUs themselves were far less efficient than those available now.
The target audience probably also played a part, at least for the first PCs. Contemporaneous home micros had no fans and were silent (barring electronic and drive noise); but PCs were office tools first and foremost, and noise was less of a factor. I think it’s also fair to say that IBM had a tendency to err on the side of caution, favouring sturdier components and designs even if it meant they were noisier. (Later on there were fanless clones, such as the Sinclair PC200.) The push to make cheaper computers, later in the 80s and especially early 90s, didn’t help with noise: cheaply-built and assembled systems often had quite a lot of case buzz, and electronic noise from cheap components wasn’t unusual.
For quite a long time too, fan noise was a minor component in the overall picture: fan noise was constant, as was the whine from hard drives, so it ended up fading into the background, unlike the constant chatter of floppy drives, hard drives seeks, and for those of us unlucky enough to have to print regularly, dot-matrix printers. As far as I remember, in the PC space fan noise became a concern for manufacturers in the late 90s and early 2000s, especially in the race to GigaHertz which led to small, noisy CPU fans and many fried CPUs.
All that racket died down as components improved: floppy drives disappeared, hard drives became quieter (this was an important feature in the early 2000s), cases became quieter, heatsinks got bigger, fans improved and acquired the ability to vary their speed as necessary, and CPUs started taking their thermal envelope into account.
The two computers you mention were presumably representative of their eras. The 1992 PC came just before manufacturers started paying much attention to energy consumption (the Energy Star program started in 1992) or even noise; it would likely have a somewhat noisy PSU, and a whiny hard drive. The late 90s PC came after; it would have benefited (slightly) from improved CPUs with varying power consumption, quieter hard drives, and likely had more than one fan, leading to a lower noise floor than the 1992 PC.
Don’t read too much into CPU evolution though in the context of this question. Microprocessors in the 80s and early 90s didn’t produce enough heat to pose any threat to themselves; however their energy consumption and heat production didn’t drop when they were idle (at least, in desktops). Starting in the mid-90s, x86 CPUs did reduce their heat production when idle, but their thermal envelopes increased dramatically as Intel and AMD raced to produce the first GHz CPU. Since then, idle power consumption has decreased an awful lot so that an idle CPU now doesn’t take much energy at all.