Guitarpima
If you are a child, you may hear a difference between the different sample rates but if you think you can as an adult, your kidding yourself.

AT
Like most attempts at high end, if there is a difference between 44.1 and 96 or 768 it can only be heard on a good system and in good rooms by trained ears.  Just like converters and mics and preamps and such.  It is a small diff that most people don't get because they simply can't hear it.  Musicians and engineers etc. w/ ears attuned to the nth degree can tell some differences, but I imagine would be hard pressed if blind tested on ear buds.
 
Of course, many of us get our rocks off on the last inch of quality we can squeeze out of sound whether by hardware or technique, and I warrant many of us get better performances and give greater detail to our work if we believe we can hear such differences.  So it is worthwhile, even if only psychologically.  But that doesn't mean you have to go all DSD or only use the most exotic hardware etc.  The best bet is to do testing yourself on your system and see if you can hear the difference, then decide if filling up hard drives like they are floppies is worth it to you.
 
Me, I record at 44.1 - I haven't heard enough difference to switch to 96 even tho it is becoming standard.  I do make sure the effects upsample and turn the older Cake synths etc. to high render rates.  That I can hear a difference in the smoothness.
 
@

drewfx1

lawp
as craig points out, it's all in the maths (whether you can hear it or not ;-)) i.e., generally, the bigger the digital-unit-to-realworld-unit ratio, the better or more accurate it all is

No.
 
Once you are more accurate than human hearing, you cannot get better accuracy unless you replace the human with something better. 

Milton
Here's how I perceive different sample rates:  44.1 = Mp3, ear buds, iTunes, iPod, average car stereos, boomboxes     96KHz = CD, good studio monitors, and above average home stereos     192KHz = audiophile home stereos (and mostly classical and HD recorded acoustic music). So the final medium and intended audience has a factor in the sample rate choice for "capture". It's similar to photography and pixels:  5 mp camera makes an excellent photo but not past a 5x7 print    8mp camera makes excellent 8x10 inch print    20mp camera makes excellent 16x20 print etc. But if one is only going to print 5x7 photos then having a 20mp camera is then a waste. My intended audience is CD and above average home stereos. But downsampling from 96 to 44.1 can let me print "any" size photo from 5x7 to 16x20 (figuratively). eg. Mp3, earbuds to higher end stereo systems. If my theory is incorrect, please enlighten me.

Milton
Here's how I perceive different sample rates:  44.1 = Mp3, ear buds, iTunes, iPod, average car stereos, boomboxes     96KHz = CD, good studio monitors, and above average home stereos     192KHz = audiophile home stereos (and mostly classical and HD recorded acoustic music). So the final medium and intended audience has a factor in the sample rate choice for "capture". It's similar to photography and pixels:  5 mp camera makes an excellent photo but not past a 5x7 print    8mp camera makes excellent 8x10 inch print    20mp camera makes excellent 16x20 print etc. But if one is only going to print 5x7 photos then having a 20mp camera is then a waste. My intended audience is CD and above average home stereos. But downsampling from 96 to 44.1 can let me print "any" size photo from 5x7 to 16x20 (figuratively). eg. Mp3, earbuds to higher end stereo systems. If my theory is incorrect, please enlighten me.

lawp
as craig points out, it's all in the maths (whether you can hear it or not ;-)) i.e., generally, the bigger the digital-unit-to-realworld-unit ratio, the better or more accurate it all is

drewfx1
Sampling is nothing like photography.

drewfx1
Accuracy is generally always limited by the "least accurate" step. IOW, a noisy signal does not become "more accurate" by doing calculations on it at higher precision once any calculation errors are already buried in the noise present.