Why dont you want a microphone to replicate as nearly as possible in the recording what is heard acoustically, if such a mic would give more versatility and freedom?
Those are good questions. Basically, it depends on what you're doing.... I'll answer the question more thoroughly at the bottom after babbling for a bit.
Instead of having a mic that has a certain coloration, why can't you add the coloration with EQ later on?
In theory, you can. In practice, mic modeling is really complex and never completely captures the character of a particular mic. It requires a lot more than EQ to get in the ballpark.
The first things that come to mind are:
1. nonlinearity. Some mics contain caps that are inherently nonlinear. In those cases, the level of near-DC in the input signal can have an effect on the frequency response, depending on how the part is used in the circuit. And it's not just caps that behave nonlinearly. Tubes can behave nonlinearly, as can the capsules themselves, but I'll talk about that later.
The general result when you get this behavior is that a big low frequency component can cause the high end to become less pronounced or more pronounced. That's really hard to reproduce with EQ.... This can also (in extreme cases) result in a brittle high end that is basically impossible to fix with EQ (short of removing the high frequencies entirely).
2. off-axis response. Different mics exhibit differing response off axis. This can cause you to get different frequency response of reflected sound, not just volume. This has some effect on the overall sound.
3. transient response. Some types of mic capsule have slower transient response than others. While this may look on the face like high frequency loss, in reality, this is a form of nonlinearity creeping in again. Loud high frequencies get damped more than quiet ones. In general, the slowest transient response comes from dynamic mics (with the largest diameter dynamics being slowest), followed by condensers (with large diaphragm condensers being slowest among them), followed by ribbons (with thicker ribbons being slowest, I think???).
4. slew rate. This is basically like transient response, but coming from the circuitry in the microphone.
5. ringing and added harmonics. Cheap transformers ring, producing distortion that is hard to fix (the insertion of additional high frequency content immediately after a transient). Better transformers add harmonics to the signal without ringing uncontrollably. This may or may not be pleasant, depending on your taste, the sound source, etc. Either way, it's hard to replicate outside of hardware.
Ringing:
6. phase response. I'll mostly leave this for somebody else to explain, as I don't fully get it. I -think- the short explanation is that you can end up with slight variations in the timing of signals at different frequencies. This can increase or decrease the harmonic complexity in the resulting signal, depending on how those delayed high frequencies line up with harmonics of the lower frequency signals that get added later due to #5.
That said, the short answer to your question is this: if you pick a microphone that has a pleasant effect for a particular source, it is less work later. Picking a flat measurement mic or whatever is good for some sources, but most sources sound better with a little tweaking, and voices tend to sound a lot better if you reduce the transient response. How's that for a short answer?
