Just a quick interjection, there are other factors that will "color" a sound that have nothing to do with the converters. The op-amp (Operational Amps) are probably the main culprit in the input stage that will lead to a noticeable alteration from the source (perceived as a EQ differences and variations in the harmonics of the original source.
The better the interface, the better the op amps. "Harsh" sound is usually perceived from a 'low quality" input stage that has already thinned out some frequency ranges (before you even get to the A/D stage).
Another factor, a common misconception, is the someone will say that the interface "colors" the sound and that it is "muddier" when in fact this is just a shitty internal clock (as discussed before in other threads) and is not a colorization at all... it is merely a high jitter ratio on the captured samples and thus the sound is perceived as "thin" or "colored" or "muddy" due to variations in the sample clock crystal (NOT the A/D at all)
Let's also remember, the D/A also has Op amps (separate from the input stage) and if you are monitoring via your interface, some colorization will take place at the output stage. A test I use on other peoples tracks, is to listen to the tracks back via SPDIF (lightpipe) thru a few interfaces to get a better idea of the sound of the actual tracks. By doing this, you can at least go digital directly to a few different monitor sources, without added coloring.
I would say that the 'warmth' I heard in old Apogee interfaces were a combination of great Op amps, good converters and a solid external clock. So when I think back now, the sound wasn't "colored" to sound warmer, it just sounded better because the smearing, and muddiness was not present... so in fact, it was just that it sounded closer to the source, not that it was colored.