I'm a photographer and I use the histogram on the back of my camera all the time to see if I've blown out the highlights. Reason being, sometimes you don't notice until a shoot is over that there is one section completely lost. Problem with this analogy is that Photography IS visual, so visual clues have a direct bearing. Music is audible, so translations of that into visual clues has a different function.
There's another important difference. When you're looking at the on-camera histogram, that's because you don't have the image proper to look at, either on a calibrated video screen or in a test print. Because you can't yet see the actual product, you use the histogram for on-site help. That's fine.
When working on music post, however, one has the music playing on their studio monitors right in front of them. It's not like they need to look at the graph because they can't hear the actual product because they're not near some loudspeakers. In fact, it's just the opposite, because they do have their studio monitors in front of them, the analyzer should not be necessary.
Spectral Analyzers are good for us less experienced engineers. It is a great starting point to see what we may not be able to hear at this point. For example, someone teaching themself at home may have no clue what 180 hz sounds like versus 500 hz. they may just know the song sounds muddy. "Seeing" is a great way to help learn the differences.You can see a spike around 400-700 hz (for example) and after a few mixes you now know what frequencies are causing what sounds.
Unfortunately there's no way of knowing without having the proper ears just by looking at a spectrum if any given spike in it is a legitimate one or not; not all spikes are bad. And conversely, mud does not always ID itself as something that sticks out on a spectral graph; as often as not it's just an innocuous looking part of the average amplitude level.
This in fact, after thinking about it, is a problem with the theory of savants who can hear the music in their head just by looking at a spectral graph; it's entirely possible for two completey different sounds to have practically (if not necessarily perfectly) identical graphs. When you see a spike at around 500 Hz, does that mean that there is one or more instruments playing a B4 or a subharmonic of that note, does it represent a natural beat frequency "phantom note" caused by the mixing of various notes or tones, or does that mean that you're getting an excess of resonance or other kinds of mud there? You can't tell by look at a graph just what you're hearing.
And when you hear mud, does that mean that the mud is actually around that peak you see at 500Hz, or could it possibly actually be part of the general volume level with no real spikes down around 200Hz? Until or unless you can tell with your ears what 250Hz and 500Hz actually sound like, or at least get close enough for horseshoes, you have no dependable way to actually interpret what you're seeing.
Tell me, when you look at a spectrogram of a song, how can you tell if the song looks the way it's "supposed to" or not? You can't. No one can. Because there is no "supposed to".
Using a spectral analysis to learn what one is hearing is kind of putting the cart before the horse. But of more importance, it's doing it the hard way, because it'll lie to you half the time. It'll make you think that what you're looking at in the graph is what you think you're hearing, but as often as not it won't be - at least not until you have the ears to discriminate that for you.
If one wants to learn how frequencies sound, far better that what they do is sit down with a range of music styles on CD and play them through a 12-18 band graphic EQ with all the bands turned down. Then as you play the music, take one band at a time and turn it up and listen to how that sounds. How does it affect the various instruments playing at the time and what does that band sound like against each of those instruments. Then turn that band back down and raise the next one, and listen some more. And so forth down the line.
Do this for an hour a day for just a week or two, then have a friend or family member do the same thing for you, except with different CDs that you weren't practicing with, and miving random frequency bands in any order they wish, with you not looking at what he's turning up and down. Then you have to try and guess which bands they are moving up.
You may not get them all exactly correct, and that's OK, as long as you can get at least close most of the time. It really does not take long to get the hang of it; start now and you can be pretty proficient at it by Valentine's Day.
Then you will be ready to actually start mixing and to do so without having to see what you hear.
As for recommendations for plugs, I already made mine. Voxengo Span for FFT spectral analysis. I do not recommend using a combined EQ/analyzer, as most analyzer-identifiable issues are not best addressed by EQ.
Though there are some few exceptions; a harmonic noise series can best be addressed via an EQ with a harmonic filter, for example. But the only plug I know of with a built-in harmonic filter and a spectral analysis in one plug is the latest version of Roger Nichols' Frequalizer, which is now out of print.
G.