Ok, so let me start with a a qualifier: there's no one way to describe a scene. I'm going to be providing instructions for _a_ way to do so in the hopes that they will be useful to anyone who is looking for some guidelines in stepping away from boxed text. The ultimate goal of such guidelines is to become unnecessary. The upshot of this is that if you have other tricks you want to suggest, please feel free to do so in the comments. This is NOT a comprehensive approach, and more tricks are always useful. Now, with that out of the way...
Let me start without he assumption that you need to describe a scene to your players, and that you need to create this description from whole cloth. You may have boxed text you don't want to read, or you may have a description that you need to paraphrase. The goal is to turn that into a description that actually works for your game.
For your description, you need to provide four things, the FRAME, the THREATS, the OPPORTUNITIES and the DETAILS. If you can cover each of those points quickly and clearly, you'll have a solid description.
The Frame is the first part of any description, and in some ways the most important, as it establishes the baseline. Since you can't describe everything in detail, the frame helps set the tone for the details you don't provide, helping your players fill in the gaps. You should consider it a sketch, and you want to answer these question: What is the approximate size and shape of the space? Is there anything peculiar about the arrangement of the space? Where are the characters within the space?
You should be able to answer those questions in one or two sentences. For example:
You are on a crowded city street that serves as an impromptu bazaar. It's easily forty feet wide, and goes fifty feet in each direction before hitting a cross street.
Old, squat trees press up against the right side of the trail, blocking sight and movement. To the left, the hillside falls away in a steep embankment that offers a lovely view and the potential for an unpleasant drop.
The alley is cramped, narrow enough that you can touch both sides with outstretched arms. The street is right behind you, and about 15 feet in the alley turns left and out of sight.
The door you came through opens into the center of one side of this rectangular room - the side you emerge from is fifty feet wide, and it is perhaps thirty feet across to the opposite wall.
By itself, the frame doesn't tell is much about the scene - specifically, it doesn't tell us why the scene is interesting, and why we're describing this particular scene rather than some other one. That's where THREATS and OPPORTUNITIES come up.
THREATS are usually pretty obvious, and while it might be easy to generalize them as creatures or traps, the real thing about threats is that they demand player action (even if that action is to sneak away quietly). Threats are not generic - simply saying there's a monster in a room doesn't convey the threat because it does not lead to action. Is the monster sleeping? Distracted? Charging at you? Loading a crossbow? What the creature is doing (and how) is as important as what it is. This is why certain information (such as what weapons an enemy has) is part of the threat - it suggests what action the enemy is going to take.
When the threats aren't creatures, the same rules apply - just describing the existence of the threat is not going to tell your players enough. If there's a pendulum blade scything across the room, it's very important to know whether it's scything _towards_ the characters.
Thus, with the threat, you must answer the following questions: What is the threat, what is it doing and (if it's not obvious) what are the consequences of it doing that? For example:
A centaur on the far side of the room startles as you enter, instinctively drawing an arrow from the quiver on his back and scanning you for the optimal target.
The door slams itself shut behind you with a *click* and a small hole, the size of a mans fist appears in the center of the room, though after a few seconds, it has grown to the size of a dinner plater, and is continuing to grow.
Argale waves to you from across the plaza, seemingly oblivious to the men in crimson cloaks approaching him from the opposite side. They're closer to him than you are, and are holding their hands under their cloaks in a way that strongly suggests hidden weapons.
OPPORTUNITIES are the flip side of threats. They are things which players may use, exploit or take advantage of but which do not necessarily have any impetus of their own. They are the things which, if you were to see in a movie, you just _know_ are going to get jumped on, knocked over, blown up, swung from or otherwise used to liven up the scene. As a GM this requires a little bit of thought. Some elements (like neutral "traps") may be a mechanical part of the encounter, but other things may be a little more neutral. As a good rule of thumb, opportunities include anything which it has occurred to you the players might use or interact with in the scene and which you're ready to handle.
Opportunities answer the question "What things might the players use?"
A great iron chandelier hangs from the ceiling, with the winch and chain holding it attached to the far wall.
Heavy tapestries hang from each of the balconies, almost down to the floor.
The ancient statue of Thor up against the wall has seen better days; it's a wonder it's still standing.
DETAILS are the elusive "everything else" and they're where you're most likely to trip up in providing a description by providing too many details. The temptation is to treat details as their own thing, and to launch into a thorough rundown of the contents of the scene in meticulous detail. Don't.
The other temptations to use details as an outlet for your internal author, busting out vivid, moving descriptions of people and things which will totally suck your players into the moment. That doesn't work either, and in fact one reason so much boxed text is so bad is that it strives to do just that.
The trick is that all the details you need should be included in the FRAME, THREATS or OPPORTUNITIES. Look back at the examples I've given so far and imagine if I'd been even more terse: A tall crumbling statue is up against the wall. The street is an impromptu bazaar. A hole has appeared in the middle of the room and it's growing.
All of those would have conveyed the critical details, but nothing more than that. By including a tiny bit more information in the form of small details, like that it's a statue of Thor, or the merchant's blankets, I suggest a great deal more.
And that's the trick. A good detail says a lot about the scene without explicitly stating it outright. It is a shorthand for something that would take much longer to describe. Consider the following:
You are in an opulently furnished bedroom.
That gives us some information, but how do you fill in the details? How big is the bed? What shape is it? Other furnishings? Windows? Art? You need to give the description some sort of touchpoint. For example:
The bed is a giant four-poster monstrosity, carved from the same severe, dark wood as the rest of the furnishings.
The bed is a circular cushion on the floor directly under the skylight, covered with silken pillows and surrounded by a gauzy curtain.
Either one of those would be a valid way to provide some detail about the room but they suggest _drastically_ different bedrooms. And with only one element detailed (the bed), how many other things have been suggested about the room?
The trick is that any time you feel you have a lot you need to say, pick one element of it that seems reasonably iconic. The art on the walls. The style of dress. The smell of the place. It will not convey everything you might have in your mind, but it's a function of diminishing returns. That first detail captures ~75% of the flavor of a scene. The next one only brings it to ~80%, and each subsequent detail adds a little less. The number of details you need to add to get near 100% is enough to be cumbersome. If, instead, you embrace the approximate nature of key details, you can get onto playing more quickly.
Ok, so now comes the time to tie it all together. While I would not suggest reading pre-written text, I might suggest it's worth writing down you description as you intend to provide it. The purpose of doing so is not to lock you into those specific words, but rather to plant the key points in your mind.
As a rule of thumb, try to keep the description under 100 words, which will end up being a little over 30 seconds of speaking time. That's short enough to keep your players focused, but long enough to include some details. If you're describing a whole new place for the first time then you might go a little longer, but err on the conservative side.
Try to shoot for 7 sentences, 2 sentences of frame, 2 sentence of opportunities and 2 sentences of threat, with one free-safety sentence to use as you see fit.
That is all a lot of verbiage for something that boils down to 7 sentences. But some of the simplest things are the hardest to explain.
1 - Part of the assumption is that the information you need to describe has not been presented to you in an immediately useful way. There are means (both hypothetical and real) of presenting that data more usefully, but we won't get into that here.
2 - Describing where things are, especially characters, can be complicated, especially in oddly shaped spaces. If you're using a battle mat, you're saved a lot of this hassle, but if you're not, consider a quick sketch as the solution to something too complicated to explain.
3 - The threat can most often be followed by the question "What do you do?" which is why it is usually the last part of any description.
4 - But be aware there's no guarantee they will ever use them. Also, consider how your opportunities and threats interact. If one of the opportunities in the scene is an awesome elevated nook where an archer can fire from cover, then there's a good chance the enemy has already places and archer up there.
5 - This is a bad habit left over format he early days of dungeon crawling when the description of a room was often a puzzle of its own. Players were expected to engage every detail listed because if they overlooked one, odds were good it was a polymorphed purple worm or the key to the incredibly important secret door opened by 1970's rock opera lyrics. This oversight would usually mean they got their faces eaten. If you're playing a game in that mode, then you're better off just sticking to the boxed text anyway because it's less likely that you'll "give something away" with your inflection.
6 - In this case the bed is part of the frame, unless you're playing the sort of game where it's the opportunity.
7 - This is where the basic creative writing teacher speaks up to remind you that you don't have to just use sight. Sound and smell are powerful descriptors, and you should use them, but sparingly. They are powerful because they're anomalous - our attention will be drawn to any scent or sound described unless they're _always_ described. So don't use them just to jazz things up, use them to draw attention to specific things.
8 - And if there's confusion, then your players can ask questions. Yes, answering questions is less efficient than describing everything up front, but it's more efficient then trying to answer every possible question before it's asked. Also, questions are useful. When a player asks a question, first ask yourself why you didn't mention that detail. If the answer is "Because it's a niggling thing" then you're ok. As an additional trick, when your player's ask a broad question that is full of stupid details like "What are the contents of the desk?" and you've already mentioned everything important, then it's reasonable to ask "What are you looking for?"