I had a 5.25" hard drive in its own enclosure. The 2-inch fan was loud enough to remind me of a jet engine. The helpline said the fan had to run at high speed so it could reduce temperature to specs. I decided to quiet it down anyway. I installed a resistor inline. It reduced rpm. It also reduced cooling effect but I decided I could live with it.
To create airflow under the disk, I raised it up on homemade standoffs. The drive gave me no trouble for the years I owned it.
It occurred to me that I ought to install some kind of thermometer inside the enclosure. It would have been fun for me to construct an electronic circuit around a Radio Shack temperature sensor. But I didn't bother, as long as the drive worked okay.
Active noise cancellation:
I doubt that this guves goid results, because:
* most of the noise is really "random" noise.
* the source of the generating nouse is relatively big...and moving.
If the fans are not internal coolers, perhaps adding external dust filter likewise these used on industrial cabinets could reduce a few, but I'm not so sure of the result.
Is it 48V/2.5A each? If so, you are using high speed fans which are producing quite high noise sound level (e.g. 75 - 85 dBA) by design. If the high airflow is actually required by the application, there's little change to reduce the noise level without completely redesigning the cooling system. Temperature or power demand controlled fan speed may be an option to reduce at least the standby noise.
Acoustic insulation might be an option if you have plenty of room.
As mentioned above, by redesigning the whole (mechanical) system could achieve some result, as for example by increasing the heatsink area could reduce the airflow requirements.
Active noise cancellation:
I doubt that this guves goid results, because:
* most of the noise is really "random" noise.
* the source of the generating nouse is relatively big...and moving.
I don't know that it won't work. The principle is that you are adding noise of opposite polarity to the source so that the sum is a net cancellation. I don't think it matters if it's random or not. But I've been wrong before.
The OP was asking for an acoustic solution. Water cooling is not an acoustic solution; however ear plugs would be.
A general problem is that non-periodical (=random) noise above a certain frequency related to source dimensions and sound velocity can be neither canceled by feedback nor by signal prediction.